Moving from the old Taostats API to the new one

This guide is for anyone, person or AI agent, whose code calls the old Taostats API. Read the first three sections, change your code, then use the index to check each endpoint you call.

[!WARNING] The old /api/ endpoints switch off on 9 November 2026. After that date, only /v1/ answers. Your API key works on both, so you can migrate and test side by side until then.

Each mapping in this guide was checked by calling both APIs for the same subnet, account or block and comparing the answers field by field.

The version comes first, there is no /api prefix, and names are plural. Each endpoint accepts its own list of sort columns, often shorter than before. For example, /v1/blocks sorts by timestamp only. The index gives the new path for every old one.

The default page size is now 50, and asking for more than 200 is an error. Some old endpoints returned a whole subnet, or every subnet, in one answer when you left out limit. The new API does not, so follow next_page until it is null:

Paging stops at the millionth row: the API serves a page only while (page - 1) × limit is below 1,000,000. To read further back, narrow the request with block_start/block_end or timestamp_start/timestamp_end instead of asking for a later page.

Balances, prices and other values that do not fit safely in a JSON number arrive as strings, in RAO where they are amounts. One TAO is 10⁹ RAO.

Addresses are plain SS58 strings now. Where the old API returned { "ss58": "5F...", "hex": "0x..." }, read the string directly.

The old API ignored mistakes. The new one rejects them, and the error message tells you how to fix the request:

Run your test suite, or each call you make, against /v1/ and work through the errors. Most of them are fixed in one line.

[!TIP] Getting a 403 from Python's built-in urllib? Its default User-Agent is blocked. Send any User-Agent header, for example my-app/1.0. requests, curl and fetch are not affected.

These rules apply to every endpoint, so the per-endpoint entries below do not repeat them. Where an endpoint breaks one, its entry says so.

The new API answers 400 to each of these. The old API ignored them or returned something.

These return 200, so nothing tells you they happened. Check them by hand.

[!WARNING] network=all now means finney only. On most endpoints all is still accepted, but it no longer returns nakamoto and kusanagi rows. If you need all three networks, ask for each one separately.

The default page size is also 50 now. Code that left out limit and expected a whole subnet (up to 1,024 rows) gets the first 50, so check next_page. And big or decimal numbers arrive as strings, such as "1234567890123": arithmetic on them without parsing will concatenate or fail.

Both APIs live at https://api.taostats.io, with the same key in the same header. List responses are still { "pagination": {...}, "data": [...] }, with the same pagination fields (current_page, per_page, total_items, total_pages, next_page, prev_page). page and limit work as before.

• CoinGecko, CoinMarketCap and TradingView endpoints keep the formats those services expect: no data/pagination wrapper, and an unknown parameter is ignored rather than rejected. TradingView history returns up to 5,000 bars in one answer.
• /v1/alpha/portfolio has no paging. Sending page or limit is a 400.
• /v1/validators/active has no paging, and /v1/miners/coldkey-summary takes no limit. Sending either is a 400.
• Basket figures (nav_per_share, rate, performance, the returns, twr) and the validator yield APY figures are JSON numbers, not strings. The basket field twr_first_block is a string.
• POST /v1/validators/yield takes order_by and order_dir in the JSON body, not the query string.
• takes no or ; it always sorts by alpha, largest first.

38 old endpoints have no direct replacement. Find yours below. Where the data still exists, the right-hand column says where to get it. Each endpoint's full entry is further down.

173 old endpoints: 114 replaced, 21 partly replaced, 38 with no replacement. Each one has its own entry further down.

One entry per old endpoint, grouped by area. Search this page for your old path.

Replaced by: GET /v1/accounts/{address} for one account, and GET /v1/accounts/leaderboard for a list of accounts.

Parameters: To look up one account, put the address in the path: /v1/accounts/{address}. The leaderboard has no address filter. On the leaderboard, the balance range filters (balance_free_min, balance_total_max and the rest) take whole numbers in RAO. The sort columns are unchanged.

Response: /v1/accounts/{address} returns a single object. balance_liquidity and balance_liquidity_24hr_ago have been removed from both routes. Leaderboard rows do not include the *_24hr_ago fields, alpha_balances or coldkey_swap. The old API returned those as null on lists anyway. Timestamps include milliseconds.

Replaced by: GET /v1/accounts/{address}/history

Parameters: address moves into the path. order_by accepts only timestamp. Every other filter is unchanged. As before, all networks are returned unless you pass network.

Response: The balance fields are strings in RAO. The old API returned them as JSON numbers on this route. balance_liquidity and coldkey_swap have been removed. Timestamps include milliseconds.

Differences: For coldkey 5HRPxma1TuTY4gPdAs4bPgyTcen13P43rucRiHppV6geJrqM, both APIs return 1,930 rows with the same balances. Three fields differ. At the same block, rank is 81 on the old API and 80 on the new one. root_basket_claimable_tao is null on the old API and filled in on the new one. On kusanagi rows from 2021, root_claim_type is IGNORE on the old API and null on the new one.

Replaced by: GET /v1/tax/report

Parameters: unchanged.

Response: unchanged. The whole report comes back on one page. Checked on one coldkey for one day: 327 rows on both APIs, with the same fields and the same values.

Replaced by: GET /v1/tax/report?format=csv

Parameters: Add format=csv. The other parameters are unchanged.

Response: The header is the same, and both APIs return the same 328 lines. Rows that fall in the same block may come back in a different order.

Replaced by: GET /v1/tax/active-tokens

Parameters: unchanged.

Response: The same tokens come back. Subnet tokens are listed in netuid order.

No replacement. The new API has no combined coldkey report. You can rebuild its daily balance rows from two routes:

• GET /v1/accounts/{address}/history?network=finney gives the daily balances. Its balance_total, balance_free and balance_staked are in RAO, where the old report gave TAO. For example, balance_total 109993431919652 is the old total_balance 109993.431919652.
• GET /v1/price/ohlc?asset=TAO&period=1d gives a TAO price for tao_price.

The report's transaction rows (transaction_type, debit_amount, credit_amount) are only available one token at a time, from GET /v1/tax/report. To get every token, call GET /v1/tax/active-tokens first. daily_income and daily_income_usd have no equivalent.

No replacement. This was the CSV form of /api/accounting/coldkey_report/v1. The same rebuild applies. For the transaction rows, GET /v1/tax/report?format=csv gives CSV for one token at a time.

Partly replaced by: GET /v1/subnets/neuron-registration-events

Parameters: coldkey is unchanged. Replace date_start and date_end with timestamp_start and timestamp_end in Unix seconds.

Response: The rows in the old neuron_registrations array are now the rows of this list. hotkey and coldkey are plain SS58 strings. The new API has no neuron_registration_cost; add up registration_cost across the rows instead. OLD fields income and stake_balances have no new equivalent. In every sample we took, the old API returned income as "0" and stake_balances as empty.

Replaced by: GET /v1/blocks

Parameters: The spec_version and validator filters have been removed. order_by accepts only timestamp. network no longer accepts testnet.

Response: The fields are unchanged. Timestamps include milliseconds.

Replaced by: GET /v1/tokenomics/emission

Parameters: order_by accepts only timestamp.

Response: The fields and values are unchanged on the blocks we checked. Timestamps include milliseconds.

Differences: History on the new API starts at block 1,404,225. On the old API it starts at block 0. The new API has 7,827,980 rows and the old API has 9,232,204.

Partly replaced by: GET /v1/subnets/pools/total-price/history

Parameters: The parameters are unchanged, including frequency, which also accepts by_block. order=date_asc and order=date_desc become order_dir.

Response: Each row is the same block as on the old API. NEW fields price, alpha_volume, alpha_buy_volume, alpha_sell_volume, root_volume, root_buy_volume and root_sell_volume are new.

Differences: Data starts on 2025-02-13, at block 4,921,036. Earlier windows return no rows: a window in January 2024 returned 3 rows on the old API and 0 on the new one. A block that falls exactly on timestamp_end at midnight is included as an extra row. One three-day window returned 4 rows on the new API and 3 on the old.

The new API writes args differently from the old API. This applies to calls and extrinsics, to the call inside a proxy call, and to the args of events:

• Keys are snake_case, for example amount_staked where the old API had amountStaked.
• An account is {"Id": "5..."}, an SS58 address, where the old API had {"__kind": "Id", "value": "0x..."}.
• A whole number is a JSON number. It becomes a string only if it is larger than 9007199254740991. The old API always used strings.
• An enum is its variant name, for example "Normal", where the old API had {"__kind": "Normal"}.
• An optional argument is "None" or {"Some": value}. The old API left the key out, or gave the bare value.
• A nested call is {"Pallet": {"call_name": {...}}}. The old API used {"__kind": "Pallet", "value": {"__kind": "call_name", ...}}.

Replaced by: GET /v1/calls

Parameters: NEW parameters pallet, name, signer_address, include_args and fields are new. full_name still works. network no longer accepts testnet.

Response: args is left out unless you pass include_args=true. full_name has been removed; it was always pallet + . + name. origin_address is now an SS58 address; on the old API it was hex. NEW fields signer_address and args_summary are new. Timestamps include milliseconds.

Differences: origin is filled in only on the top-level call of an extrinsic, and is null on the calls nested under it. The old API filled it in on nested calls too. On extrinsic 9000001-0009, both APIs return the same 3 calls with the same ids.

Replaced by: GET /v1/events

Parameters: order_by accepts only timestamp. The old sorts on phase, pallet, name, id and extrinsic_id have been removed. network no longer accepts testnet.

Response: full_name has been removed; use pallet and name. extrinsic_index has been removed; it is the number after the dash in extrinsic_id. args uses the new encoding described above, for example "weight": {"proof_size": 26563, "ref_time": 2886592370}. Timestamps include milliseconds.

Differences: call_id is filled in. On the old API it was null. For example, event 9000001-0166 has call_id 9000001-0009 on the new API. Both APIs return the same 11 events for extrinsic 9000001-0009.

Replaced by: GET /v1/extrinsics

Parameters: NEW parameters pallet, name, include_args and fields are new. order_by accepts only timestamp. The old sorts on id, success and signer_address have been removed. network no longer accepts testnet.

Response:

• call_args is now args, and it is left out unless you pass include_args=true.
• full_name has been removed; use pallet and name.
• version and call_id have been removed. On the extrinsics we checked, call_id was equal to id.
• signer_address is an SS58 address; on the old API it was hex.
• signature is now the hex-encoded signature string, prefixed with its signature-type byte. The old API returned an object with address, signature and signedExtensions, which held the nonce and the mortality. Those signed-extension values have no new equivalent.
• In error, OLD field extra_info is now docs, and pallet is capitalised as the chain names it (Proxy, where the old API had proxy).
• NEW fields and are new. is the account that paid the fee.

Replaced by: GET /v1/proxy-calls

Parameters: order_by accepts only timestamp. The block sort has been removed.

Response: args uses the new encoding described under "Calls, events and extrinsics", for example {"SubtensorModule": {"add_stake": {...}}}. In this route's args, some number arguments come back wrapped in a one-item array, for example "amount_staked": [5000000000] and "netuid": [100]. Timestamps include milliseconds.

Differences: The two APIs list different proxy calls, so total_items will not match. The new API also lists proxy calls nested inside batch, sudo and multisig calls. Their id has extra parts, for example finney-9000177-0007-0-0. The new API leaves out a proxy call that the chain rejected, such as one that failed with NotProxy, because the inner call never ran. The old API lists those. In blocks 9,000,000 to 9,000,300, the old API returned 45 proxy calls and the new API returned 73. All 44 top-level calls the two lists share have the same ids. The other 29 on the new API are nested calls. The one call only the old API returned failed with NotProxy.

Replaced by: GET /v1/transfers

Parameters: You can no longer sort by block_number; use order_by=timestamp or order_by=amount. network accepts finney, kusanagi or nakamoto. all is accepted but returns finney only, as on other endpoints.

Response: unchanged, apart from from and to being plain SS58 strings. On blocks 9,000,000 to 9,000,100, both APIs return the same 103 transfers.

Replaced by: GET /v1/evm/address_from_ss58

Parameters: unchanged.

Response: unchanged: a plain JSON string with the EVM address. An account with no known EVM address still returns 404. The new API also has GET /v1/evm/conversions/ss58-to-address, which converts any SS58 address to its EVM address by calculation, without a lookup.

Replaced by: GET /v1/evm/blocks

Parameters: unchanged. order_by accepts block_number or timestamp.

Response: unchanged.

Replaced by: GET /v1/evm/contracts

Parameters: order_by accepts only timestamp. You can no longer sort by block.

Response: unchanged.

Replaced by: GET /v1/evm/logs

Parameters: unchanged. order_by accepts id, block_number or timestamp.

Response: unchanged.

Replaced by: GET /v1/evm/transactions

Parameters: unchanged. order_by accepts block_number or timestamp.

Response: unchanged.

Partly replaced by: GET /v1/evm/contracts

Parameters: Only address is kept. The name and symbol filters have been removed, and there is no filter for ERC-20 contracts only: page through the list and keep rows with erc20: true. order_by accepts only timestamp.

Response: name, symbol, decimals, created_by and transaction_hash are present. OLD field created_at_block_number is now block_number, and created_at_timestamp is now timestamp. The values match on the same token. Each row also carries the other contract fields, such as erc20, erc721 and owner.

Partly replaced by: GET /v1/evm/logs?event_name=Transfer&address=<token address>

Parameters: Filter by token with address, by transaction with transaction_hash, and by block or time range as before. The from, to, token_name, token_symbol, amount_min and amount_max filters have no equivalent. order_by accepts id, block_number or timestamp.

Response: Each transfer is a log row. OLD field amount is args.value, and from and to are args.from and args.to. The new API writes those two addresses in mixed case. The values match on the same transfer. The token is the row's address. OLD fields token_name, token_symbol and token_decimals are not on the row; look them up with GET /v1/evm/contracts?address=. Tokens of other standards emit a Transfer event too, so keep only addresses that /v1/evm/contracts marks erc20: true.

No replacement. The new API does not list ERC-20 token holders or their balances. Neither balance nor total_transfers, nor the first and last active block, is available.

Replaced by: GET /v1/contract-events

Parameters: unchanged. order_by accepts id, block_number, timestamp or name.

Response: unchanged, including args, which keeps the old encoding.

Replaced by: GET /v1/contract-events?id={id}

Parameters: The id moves from the path into the id query parameter.

Response: The event comes back as the single row of a normal page, in data[0]. The old API returned it as an object in data. An unknown id returns an empty page with status 200. The old API returned 404.

Replaced by: GET /v1/network/runtime-version

Parameters: unchanged (none).

Response: Returns a single object. The fields and values are unchanged. Timestamps include milliseconds.

Replaced by: GET /v1/network/runtime-version/history

Parameters: order_by accepts only timestamp.

Response: unchanged. Timestamps include milliseconds.

Replaced by: GET /v1/accounts/pending-coldkey-swaps

Parameters: unchanged (none).

Response: old_coldkey is a plain SS58 string. NEW fields disputed and disputed_block_number are new; they say whether the swap has been disputed, and in which block. Timestamps include milliseconds.

Differences:

• block_number and timestamp are the block in which the swap was announced, and that block's time. On the old API they were the current block and the time of the request.
• Once the execution block has passed, predicted_execution_timestamp is that block's real time. The old API kept an estimate. For example, for the swap with execution_block_number 7,765,759, the old API says 2026-03-17T23:54:24Z and the new API says 2026-03-17T14:26:48.000Z, which is the block's real time.
• For swaps whose execution block is still in the future, the two APIs agree.
• Both APIs list the same 17 swaps.

Replaced by: GET /v1/alpha/root-claims

Parameters: order_by accepts only timestamp. The coldkey and block_number sorts have been removed.

Response: coldkey is a plain SS58 string. NEW field tao is new. It is the TAO the claim paid out, in RAO, and is null for claims before block 8,765,684. Timestamps include milliseconds.

Differences: The new API returns one row per claim. The old API merged two claims by the same coldkey in the same block into one row, and it is missing claims around blocks 8,720,000 to 8,779,999. In blocks 9,000,000 to 9,001,000, both APIs return the same 9 claims.

Replaced by: GET /v1/exchange-addresses

Parameters: unchanged (none).

Response: The same 11 entries, each with coldkey, name and icon. coldkey is a plain SS58 string.

Replaced by: GET /v1/status

Parameters: unchanged (none).

Response: ok and version are still there; version is the new API's own version number. OLD field status.timestamp has no new equivalent. NEW field timestamp is new: it is the server's current time.

Partly replaced by: POST /v1/rpc/http

Parameters: The request body is the JSON-RPC 2.0 request itself, or a batch array of requests. The old API expected {"target": ..., "request": {...}}. You can no longer choose a target. Every request goes to the finney_lite node, so archive queries are available only over the WebSocket route.

Response: The node's JSON-RPC response is returned as it is. The old API wrapped it as {"target": ..., "response": {...}}. A batch's replies can come back in a different order from the requests, so match them by id. Not checked live.

Replaced by: WS /v1/rpc/ws/{target}

Parameters: unchanged. target is finney_lite or finney_archive. The route is served over wss:// only.

Response: If the node cannot be reached, the connection is opened and then closed with code 1011. Not checked live.

No replacement. The new API has no route that encrypts a blob to a public key. The old route took pk_hex and tx_hex and returned ciphertext_hex. Not checked live.

Partly replaced by: GET /v1/subnets/metrics and GET /v1/subnets/hyperparameters, joined on netuid. Subnet activity figures (emission, registrations, owner, recycled amounts) are on /v1/subnets/metrics; the subnet's chain settings are on /v1/subnets/hyperparameters.

Parameters: netuid, page and limit work on both routes. The old API returned every subnet in one page by default; the new API returns 50 a page, so ask for limit=200 to get all subnets. The emission_asc and emission_desc orders have no equivalent: /v1/subnets/metrics sorts by netuid only, and /v1/subnets/hyperparameters takes no sort parameter.

Response:

• These old fields have new names on /v1/subnets/hyperparameters: max_neurons is max_allowed_uids, max_validators is max_allowed_validators, max_regs_per_block is max_registrations_per_block, bonds_moving_avg is bonds_moving_average, bonds_reset_on is bonds_reset_enabled, subtoken_enabled is subnet_is_active, transfer_toggle is transfers_enabled, mech_count is mechanism_count, mech_emission_split is mechanism_emission_split, and immune_owner_uids_limit is owner_immune_neuron_limit.
• yuma3_on (true or false) is replaced by yuma_version, which is 2 or 3.
• and are replaced by one field, .

Differences:

• bonds_penalty was wrong in the old API (0.00000000000000355266). The new API gives the chain's value (for example "1").
• On 6 of 129 subnets the old activity_cutoff was out of date. The new API gives the value the chain currently uses.
• burn_increase_mult, alpha_high and alpha_low are given to a different number of decimal places.

Replaced by: GET /v1/subnets/history

Parameters: block_number is gone; use block_start and block_end set to the same block. order_by accepts timestamp or block_number. netuid is still required, and frequency still takes by_block, by_hour and by_day.

Response: Each row now has 7 fields: block_number, timestamp, netuid, emission, excess_tao, neuron_registration_cost and recycled_24_hours (which can be null). The old rows carried about 75 fields, including the subnet's settings. For those, use GET /v1/subnets/hyperparameters (current values only). Where both APIs have a row for the same block, the shared values are identical.

Differences:

• The daily points fall on different blocks. The new API's by_day points are the blocks divisible by 7,200, which is about 10:02 UTC each day. The old API's daily points were the last block of each UTC day, with the current block as the newest point.
• by_block in the new API returns one row every 300 blocks, the same rows as by_hour. The old API returned every block.
• History starts on 14 February 2025 (block 4,924,800). The old API went back to 21 December 2024.

Replaced by: GET /v1/subnets/identities

Parameters: unchanged. The old API returned every subnet in one page by default; the new API returns 50 a page.

Response: unchanged.

Differences: summary, tags and twitter are always null in the new API. The old API filled them for some subnets (for example, subnet 64's twitter was @chutes_ai).

Replaced by: GET /v1/subnets/identities/history

Parameters: To sort by subnet, use order_by=netuid. The old order value was net_uid_asc or net_uid_desc.

Response: unchanged.

Differences: An empty identity field is null in the new API. The old API returned either "" or null.

No replacement. The new API has no route that gives a subnet token's name, symbol, decimals, platform, contract_address, circulating_supply, total_supply, max_supply, release_schedule, exchange_url, block_explorer or logo. The nearest data:

• The subnet's name and logo are on GET /v1/subnets/identities (subnet_name, logo_url).
• The token symbol and the subnet's total alpha are on GET /v1/subnets/pools (symbol, total_alpha). total_alpha is a different figure from the old circulating_supply.

Replaced by: GET /v1/subnets/owners

Parameters: order_by accepts timestamp only. The old API could also sort by block number, which gives the same order.

Response: unchanged.

Replaced by: GET /v1/subnets/registrations

Parameters: order_by accepts timestamp only. Sorting by registration cost (register_cost_asc, register_cost_desc) has no equivalent.

Response: recycled_at_registration has no new equivalent.

Replaced by: GET /v1/subnets/registration-cost

Response: Returns a single object.

Replaced by: GET /v1/subnets/registration-cost/history

Parameters: order_by accepts timestamp only.

Response: unchanged.

Differences:

• The new API returns one row per block. The old API returned one row per day, on the last block of each UTC day. To get one of those daily values, set block_start and block_end to that block; the value is identical.
• History starts on 13 February 2025 (block 4,920,351). The old API went back to March 2023.

Replaced by: GET /v1/subnets/neuron-registration-events

Parameters: order_by accepts timestamp only. Sorting by registration cost (registration_cost_asc, registration_cost_desc) has no equivalent.

Response: unchanged.

Replaced by: GET /v1/subnets/neuron-deregistration-events

Parameters: New coldkey filter. order_by accepts timestamp only.

Response: unchanged.

Replaced by: GET /v1/subnets/distribution/coldkey

Parameters: unchanged (netuid, required). The whole list comes back in one response. Sending page or limit is an error.

Response: unchanged. Rows are sorted by count, highest first.

Replaced by: GET /v1/subnets/distribution/incentive

Parameters: unchanged (netuid, required). The whole list comes back in one response. Sending page or limit is an error.

Response: incentive has 10 decimal places.

Replaced by: GET /v1/subnets/distribution/ip

Parameters: unchanged (netuid, required). The whole list comes back in one response. Sending page or limit is an error.

Response: unchanged.

Replaced by: GET /v1/subnets/deregistrations

Parameters: order_by accepts rank only. Sorting by netuid, registration block, immunity blocks remaining, immunity or moving price has no equivalent. The old API returned every subnet in one page by default; the new API returns 50 a page.

Response: pruning_rank is now rank.

Replaced by: GET /v1/subnets/deregistrations/history

Parameters: order_by accepts timestamp only.

Response: pruning_rank is now rank.

Replaced by: GET /v1/subnets/pools for each pool's reserves, price and market cap, and GET /v1/subnets/pools/aggregate for 24-hour trading figures, price changes and the sentiment index. Join them on netuid.

Parameters: netuid, page and limit work on both routes. The old API returned every subnet in one page by default; the new API returns 50 a page. Sort columns:

• /v1/subnets/pools sorts by netuid (the default), price, liquidity, market_cap, tao_in_pool, total_alpha, alpha_in_pool, alpha_staked or root_prop.
• /v1/subnets/pools/aggregate sorts by netuid, market_cap_change_1_day, price_change_1_hour, price_change_1_day, price_change_1_week, price_change_1_month or tao_volume_24_hr.
• The old total_tao and tao_volume_one_day orders are tao_in_pool and tao_volume_24_hr. The *_one_hour, *_one_day, *_one_week and *_one_month orders are , , and .

Response:

• total_tao is now tao_in_pool (on /v1/subnets/pools).
• fear_and_greed_index is now sentiment_index, and fear_and_greed_sentiment is now sentiment (on /v1/subnets/pools/aggregate).
• seven_day_prices, the buy, sell and total volumes, the counts of buys, sells, buyers and sellers, highest_price_24_hr, lowest_price_24_hr, last_price, the price changes and market_cap_change_1_day are all on /v1/subnets/pools/aggregate, with unchanged names.
• These old fields have no new equivalent: name, rank, fee_rate, enabled_user_liquidity, swap_v3_initialized, user_provided_tao, user_provided_alpha, protocol_provided_tao, protocol_provided_alpha, alpha_sqrt_price, , , , . The subnet's name is on ().

Differences:

• alpha_staked is lower in the new API by a fixed amount for each subnet (12,363 alpha on subnet 64, the same every day). market_cap is lower in the same proportion, because both APIs compute it as price × (alpha_in_pool + alpha_staked).
• tao_volume_24_hr_change_1_day and alpha_volume_24_hr_change_1_day can differ by a few points even when the 24-hour volumes agree. Measured on subnet 64: −29.2 in the new API and −31.7 in the old, with 24-hour TAO volume of 1,847.9 and 1,849.8.
• sentiment_index differs slightly from the old fear_and_greed_index, typically by less than half a point on the 0–100 scale.

Replaced by: GET /v1/subnets/pools and GET /v1/subnets/pools/aggregate. This old route returned the same data as GET /api/dtao/pool/latest/v1, and everything in that entry applies.

Replaced by: GET /v1/subnets/pools/history

Parameters:

• netuid is now required.
• block_number is gone; use block_start and block_end set to the same block.
• order_by accepts timestamp only. Sorting by price has no equivalent.
• frequency is unchanged.

Response: total_tao is now tao_in_pool. These old fields have no new equivalent: name, rank, fee_rate, enabled_user_liquidity, swap_v3_initialized, user_provided_tao, user_provided_alpha, protocol_provided_tao, protocol_provided_alpha, alpha_sqrt_price, current_tick, fee_global_tao, fee_global_alpha, liquidity_raw. New fields: subnet_emission_enabled, and subnet_protocol_alpha, which is always null on this route.

Differences: alpha_staked and market_cap are lower, as described under GET /api/dtao/pool/latest/v1. The daily points fall on the same blocks in both APIs.

Replaced by: GET /v1/subnets/pools/total-price

Response: Returns a single object. fear_and_greed_index is now sentiment_index, and fear_and_greed_sentiment is now sentiment.

Differences: Volume fields can differ, for the reasons given under GET /api/dtao/pool/total_price/history/v1.

Replaced by: GET /v1/subnets/pools/total-price/history

Parameters: New block_start, block_end, timestamp_start and timestamp_end filters. order_by accepts timestamp only. frequency is unchanged.

Response: fear_and_greed_index is now sentiment_index, and fear_and_greed_sentiment is now sentiment.

Differences:

• price is identical on the same block.
• The volume fields differ. From block 6,067,944, the new API does not count moving stake between hotkeys within the same subnet as trading volume. It does count the payout made when a subnet is removed as sell volume.
• On the daily row for 6 October 2026, alpha_volume was 194,743 TAO in the new API against 249,115 in the old, and root_volume was 45,488 against 35,316.

Replaced by: GET /v1/subnets/pools/total-price/history. This old route returned the same data as GET /api/dtao/pool/total_price/history/v1, and everything in that entry applies.

The new API does not serve user liquidity positions, their history or events, or the liquidity distribution of a pool. None of the five routes below has a replacement, and no field of theirs appears anywhere in the new API.

No replacement. The old route returned an empty list (checked on subnet 64).

No replacement. The newest position on the old route was opened on 22 December 2025.

No replacement.

No replacement. The old route returned no events.

No replacement. This route was a calculation, not stored data: the price for a tick is 1.0001 raised to the power of the tick (so tick 0 is price 1).

Replaced by: GET /v1/subnets/metagraph

Parameters:

• is_immunity_period is now is_immune.
• New filters: in_danger, has_dividends, has_incentive.
• order_by accepts total_alpha_stake, netuid, uid, emission, incentive, dividends, consensus and validator_trust. Sorting by updated, stake, trust, active, hotkey, coldkey, validator_permit, axon, daily_reward, registered_at or is_immunity_period has no equivalent.
• The old API returned a whole subnet (up to 1,024 neurons) in one page by default; the new API returns 50 a page.

Response:

• is_immunity_period is now is_immune.
• axon is an "ip:port" string instead of an object.
• These old fields have no new equivalent:
  • daily_reward: it equalled emission × 20 in the rows checked.
  • rank: the new miner_rank and validator_rank are different rankings.
  • trust and stake: both were "0" on every row. Use total_alpha_stake.
  • collateral: it was null on every row.
• New fields: stake_weight, hotkey_alpha, free_alpha, locked_alpha, min_locked_alpha, collateral_earned_alpha, miner_rank, , , .

Differences: total_alpha_stake is a whole number. The old API sometimes gave it with a fractional part.

Replaced by: GET /v1/subnets/metagraph/history

Parameters:

• netuid is now required.
• The hotkey and coldkey filters are gone; filter by uid instead.
• New has_incentive filter.
• order_by accepts timestamp only.

Response: The same field changes as GET /api/metagraph/latest/v1.

Differences:

• The new API returns one row per neuron each time the subnet runs its epoch (every 360 blocks on subnet 64). The old API returned one row per neuron per day, on the last block of each UTC day. That old daily row equals the new API's latest row at or before that block.
• History starts on 13 February 2025. The old API went back to 3 February 2025.

No replacement. The new API does not serve root weights or the old root fields senator, subnet_weights, rank and pruning_score. The old route had not updated since block 6,811,680 (4 November 2025), and gave stake as "0" for every neuron. The root subnet's current neurons, with their uid, hotkey, coldkey and stake, are listed by GET /v1/subnets/metagraph?netuid=0.

No replacement. The new API does not serve root weights or the old root fields senator, subnet_weights, rank and pruning_score. The old route had no rows after block 6,811,680 (4 November 2025). The root subnet's neuron history is available from GET /v1/subnets/metagraph/history?netuid=0.

Replaced by: GET /v1/subnets/metagraph

Parameters: The filters are unchanged. Sorting by hotkey or coldkey has no equivalent; the other sorts are listed under GET /api/metagraph/latest/v1.

Response:

• registration_block is now registered_at_block.
• pruning_score and trust have no new equivalent. Both were "0" on every row.
• Each row also carries the metagraph's stake and daily-reward fields, such as total_alpha_stake, alpha_stake, root_stake and daily_total_rewards_as_tao.

Replaced by: GET /v1/subnets/metagraph/history

Parameters:

• netuid is now required.
• The hotkey, coldkey, is_immune, in_danger and has_dividends filters are gone. uid and has_incentive remain.
• order_by accepts timestamp only.

Response: The same changes as GET /api/neuron/latest/v1.

Differences: Both APIs return a row per epoch, on the same blocks, with the same values. History starts on 13 February 2025. The old API went back to 3 February 2025.

Replaced by: GET /v1/subnets/metagraph/aggregate

Parameters: unchanged.

Response: The six *_pruning_score fields (max_pruning_score, max_mining_pruning_score, max_danger_pruning_score, max_immune_pruning_score, min_non_immune_pruning_score, last_dereg_pruning_score) have no new equivalent. In their place are six *_emission fields (max_emission, max_mining_emission, max_danger_emission, max_immune_emission, min_non_immune_emission, last_dereg_emission). These give emission in rao, not a pruning score, so do not compare them with the old values. The old pruning-score fields currently read "0".

Replaced by: GET /v1/subnets/metagraph/aggregate/history

Parameters: order_by accepts timestamp only.

Response: The same change as GET /api/neuron/aggregated/latest/v1: the *_pruning_score fields are gone, and *_emission fields are new.

Differences: History starts on 13 February 2025. The old API went back to 21 December 2024.

Replaced by: GET /v1/subnets/metagraph/history with has_incentive=true and timestamp_start set to the start of the period you want.

Parameters: There is no days parameter. For the last N days, set timestamp_start to now minus N × 86,400 seconds. netuid is still required. The old route returned everything in one response; the new one is paged, at up to 200 rows a page.

Response: Each row is a full metagraph row. registration_block is now registered_at_block, and incentive has 10 decimal places. Where both were checked, the rows are the same: the same neurons on the same blocks, with the same incentive.

Replaced by: GET /v1/validators

Parameters: unchanged. All 11 sort columns are kept, and the default is rank ascending.

Response: claim_types has been removed. The old API always sent an empty list. name is null when the validator has no identity.

Differences: Both APIs return 596 validators. global_nominators is higher on the new API: 41,510 against 34,887 for the same validator. The new figure matches the chain. created_on_date can be earlier on the new API, because it is the first day the hotkey held any stake (2025-02-04 against 2025-02-06 for the validator checked). dominance has 9 decimal places instead of 2.

Replaced by: GET /v1/validators/{hotkey}/history

Parameters: hotkey moves into the path and is required. The block_number filter has been removed: pass the same block as block_start and block_end instead. order_by accepts only timestamp, and the default direction is descending.

Response: claim_types has been removed.

Differences: Both APIs return 601 rows for the validator checked. The same global_nominators, created_on_date and dominance differences apply as on the latest route. The *_24_hr_change values differ slightly.

Replaced by: GET /v1/validators/active

Parameters: Only netuid is accepted. limit and page return a 400 error, and there is no paging: the whole list comes back in one response.

Response: unchanged. Rows may come back in a different order.

Differences: On subnet 1 both APIs return the same 53 validators with the same names. hotkey_alpha differs on every row, because the new API takes it from the last daily snapshot at which the validator earned. Earnings are recorded once a day, so a validator can appear or drop off up to one day later than on the old API.

Replaced by: GET /v1/validators/baskets

Parameters: order_by accepts nav_tao (the default), spot_nav_tao, shares, nav_per_share, performance, return_7d, return_30d, staker_return_7d, staker_return_30d and hotkey.

Response: nav_per_share, rate, performance, return_7d, return_30d, twr, staker_return_7d and staker_return_30d are JSON numbers. The old API sent them as strings. twr_first_block is a string; the old API sent a number. day is new. rate is never null.

Replaced by: GET /v1/validators/baskets/history

Parameters: day_start and day_end (YYYY-MM-DD) are replaced by timestamp_start and timestamp_end in Unix seconds. Each one selects the whole UTC day it falls in. order_by accepts day (the default) and the same columns as the latest route, except hotkey.

Response: The ratio fields are JSON numbers and twr_first_block is a string, as on the latest route. The current, unfinished day is not included.

Differences: Each day's row is taken at the last block of the day. The old API took it earlier, so values differ a little. For 2026-10-05 the new API's row is at block 9,220,186 (23:59:48) and the old API's at block 9,220,023 (23:27:12); return_7d is 0.00740 against 0.00875.

No replacement. No new route returns per-hotkey dividend splits (nominator_alpha_dividends, validator_alpha_dividends, the root dividend fields, nominator_return_per_kt_*, epochs or tempo). The hotkey's total stake is hotkey_alpha on GET /v1/subnets/metagraph?netuid=&hotkey=. Daily totals are daily_validating_alpha on the same route, and nominator_return_per_day and validator_return_per_day on GET /v1/validators/performance/{hotkey}.

No replacement. No new route keeps a history of the per-hotkey dividend splits. The closest history is GET /v1/validators/performance/{hotkey}/history, which has nominator_return_per_day and validator_return_per_day.

Replaced by: GET /v1/validators/performance/{hotkey}

Parameters: hotkey moves into the path. To get the same rows as the old API, pass scope=own_and_children; the default, scope=hotkey, returns a different set. The sort column v_trust is now vtrust, and type is now validator_type.

Response: parent_hotkey is new. Decimal values carry more decimal places.

Differences: alpha, take and vtrust match. nominators is higher on the new API (312 against 178 on one row). position and ratio differ, and the family_* values differ because suspended parent hotkeys are left out of the family totals.

Replaced by: GET /v1/validators/performance/{hotkey}/history

Parameters: hotkey moves into the path. Pass scope=own_and_children to get the same rows as the old API: for the validator checked that returns 1,442 rows on both APIs, while the default returns 902. The block_number filter has been removed: pass the same block as block_start and block_end. order_by accepts only timestamp, and the default direction is descending.

Response: parent_hotkey is new.

Differences: nominators is higher on the new API, as on the latest route. A child hotkey with several parents can appear under each of them.

Replaced by: GET /v1/validators/yield

Parameters: unchanged. All 7 sort columns are kept, and the default direction is descending.

Response: stake is a string. The old API sent a number. The APY fields are still JSON numbers.

Replaced by: POST /v1/validators/yield (not checked live)

Parameters: The body keeps positions (each with hotkey and netuid), min_stake, page and limit. order is replaced by order_by and order_dir. The maximum limit is 200; the old API accepted up to 1,000.

Response: As on the GET route: stake is a string.

No replacement. The new API returns only current yields, from GET /v1/validators/yield. No route keeps their history.

No replacement. This route returns a snapshot of 75 validators taken before dTAO, at block 4,920,349 (2025-02-13). The nearest data is the end-of-day stake and nominators per hotkey on GET /v1/validators/{hotkey}/history/pre-dtao, which ends on 2025-02-12.

Partly replaced by: GET /v1/validators/{hotkey}/history/pre-dtao

Parameters: hotkey moves into the path. The block_number filter has been removed. order_by accepts only timestamp.

Response: Only hotkey, block_number, timestamp, stake and nominators are kept. These old fields have no new equivalent: apr, apr_7_day_average, apr_30_day_average, blocks_until_next_reward, coldkey, created_on_date, dominance, last_reward_block, name, nominator_return_per_k, nominator_return_per_k_7_day_average, nominator_return_per_k_30_day_average, nominators_24_hr_change, pending_emission, permits, rank, registrations, stake_24_hr_change, subnet_dominance, system_stake, take, total_daily_return, validator_return and .

Differences: The new API has more early days: 9 rows against 7 for the validator checked, adding 2025-02-04 and 2025-02-05. The shared values match.

No replacement. The old route currently answers HTTP 500, so it could not be compared. The nearest data is GET /v1/accounts/identities?validator_hotkey=, which returns the identity of the coldkey that owns the hotkey, without a signature field.

Partly replaced by: GET /v1/subnets/metagraph?netuid=&hotkey=

Parameters: Filter by netuid and hotkey.

Response: active, consensus, dividends, emission, incentive, uid, updated, validator_permit and validator_trust keep their names. registered_block_number is now registered_at_block. is_immunity_period is now is_immune. These old fields have no new equivalent: daily_reward, rank, stake, trust and axon_info. The old API always sent stake and trust as 0. The metagraph route adds many new fields.

Differences: dividends and validator_trust carry fewer decimal places.

Partly replaced by: GET /v1/subnets/metagraph/history

Parameters: netuid is required. There is no hotkey or coldkey filter: look up the hotkey's uid first and filter by uid. order_by accepts only timestamp.

Response: Field names change as on the latest route. The new API stores one row per epoch (about every 99 blocks on subnet 1). The old API stored one row per day plus the latest.

Differences: The old API's history starts on 2025-03-08. Its end-of-day row at block 9,227,386 has the same emission and validator_trust as the new API's epoch row at block 9,227,315.

Partly replaced by: GET /v1/subnets/metagraph/history?netuid=&uid=

Parameters: Look up the hotkey's uid first, then filter by netuid and uid. If the uid has changed hands, keep only the rows whose hotkey is yours.

Response: blocks_since_weights_set is now updated. update_status and tempo have no new equivalent. The metagraph route adds many new fields.

Differences: The latest 4 rows match on block, emission and updated.

No replacement. The new API has no list of weight-copying validators.

Replaced by: GET /v1/validators/weights

Parameters: mechanism is new and defaults to 0. A netuid of 4096 or more is a 400 error: instead of the old combined value netuid + 4096 × mechanism, pass the subnet and the mechanism separately. For example, old netuid=4140 is new netuid=44&mechanism=1. order_by accepts netuid (the default) and uid, and the default direction is ascending.

Response: Each entry in weights carries the target's hotkey as a string. It is "unknown" when the uid has no registered neuron.

Differences: The same uids and weights come back, but the entries in weights may be in a different order.

Replaced by: GET /v1/validators/weights/history

Parameters: mechanism is new, as on the latest route. order_by accepts only timestamp.

Response: As on the latest route.

Differences: The new API keeps far more history. For the validator checked it returns 24,988 rows going back to 2025-05-09; the old API returns 2,177 rows going back to 2026-09-07.

Partly replaced by: GET /v1/validators/weights

Parameters: As for GET /api/validator/weights/latest/v2 above.

Response: call, version_key, weights_hash and reveal_round have no new equivalent. Each entry in weights carries a new hotkey field. The old API scaled the weights so the largest was 1; the new API's weights add up to 1. The proportions are the same. For example uid 248 has weight 1 on the old API and 0.6909 on the new one.

Differences: The old API returned one row per weight-setting call, so its block can differ slightly from the new API's (9,232,172 against 9,232,166 for the same set of weights).

Partly replaced by: GET /v1/validators/weights/history

Parameters: As for GET /api/validator/weights/history/v2 above.

Response: As for GET /api/validator/weights/latest/v1 above: call, version_key, weights_hash and reveal_round have no new equivalent, and the weights add up to 1 instead of having a largest value of 1.

Differences: The old API's history for the validator checked starts on 2025-10-17 (40,552 rows).

Replaced by: GET /v1/subnets/stake-events for events from block 4,920,351 on (dTAO), and GET /v1/historic/stake-events for events before it.

Parameters: nominator is now coldkey and delegate is now hotkey. action takes stake, unstake or all, instead of delegate, undelegate or all. amount_min and amount_max take whole numbers in RAO. is_transfer=false now matches every row that is not a transfer. trades_only is new: it keeps only buys and sells, leaving out stake transfers, hotkey-swap legs and, from block 6,067,944, moves within one subnet. order_by accepts only timestamp. GET /v1/historic/stake-events has no netuid, is_transfer, transfer_address, amount_min or amount_max filter.

Response: delegate is now hotkey, nominator is now coldkey and delegate_name is now hotkey_name. action is stake or unstake instead of DELEGATE or UNDELEGATE. is_transfer is always true or false, never null. registration_collateral is new: it is true when the stake is the collateral paid to register a miner. validator_swap is new: it is true when the row is one half of a move between validators within one subnet. id has a new format and does not match the old ids. alpha_price_in_tao has 9 decimal places. Rows from GET /v1/historic/stake-events have no alpha, usd, alpha_price_in_tao, alpha_price_in_usd, slippage, fee, netuid, is_transfer, transfer_address or .

Differences: For one validator on subnet 1 both APIs return 9,551 rows, with the same amount, alpha, fee and usd. slippage differs slightly on every row checked (for example 0.000106313 against 0.000100045). Before dTAO, both APIs return the same 5 rows for the validator checked.

No replacement. This route holds each coldkey's stake per hotkey before dTAO, up to block 4,920,351. GET /v1/alpha/history?coldkey=&hotkey=&netuid=0 starts at block 4,920,351, where its value matches the old route's last row. GET /v1/validators/{hotkey}/history/pre-dtao has only each validator's total stake.

No replacement. The old route currently answers HTTP 404, so it could not be compared. No new route holds stake per coldkey and hotkey from before dTAO.

Replaced by: GET /v1/validators/hotkey-family

Parameters: unchanged.

Response: stake and family_stake have been removed, from each row and from each entry in its parents and children. proportion_staked is now filled in; the old API always sent 0. Decimal values carry more decimal places.

Differences: Suspended parent hotkeys are left out of the family totals, so the family_* values can differ (for example family_root_stake 136326340507715.84 against 136326361171234). The family_* values are never negative.

Replaced by: GET /v1/validators/hotkey-family/history

Parameters: The block_number filter has been removed. order_by accepts only timestamp.

Response: As on the latest route. There is one row per day, taken at the last block of the day, and only for a hotkey with at least one parent or child. The current day is not included, and there are no rows for subnet 0. History starts on 2025-02-13.

Differences: At the same block the values match, apart from decimal places.

Replaced by: GET /v1/accounts/identities, or GET /v1/accounts/{address}/identity for one coldkey.

Parameters: address and validator_hotkey are kept on GET /v1/accounts/identities. validator_hotkey returns the coldkey that owns that hotkey. GET /v1/accounts/{address}/identity takes the coldkey in the path and answers 404 when the coldkey has no identity.

Response: The old API returned one row per coldkey and validator hotkey, plus one row with a null validator_hotkey. The new API returns one row per coldkey, with validator_hotkeys as a list. Empty strings, such as additional or github_repo, come back as null.

Differences: For the coldkey checked the old API returns 2 rows and the new API 1, with the same values.

No replacement. The new API has no history of coldkey identities. GET /v1/subnets/identities/history covers subnet identities only.

Replaced by: GET /v1/miners/autostakes

Parameters: order_by accepts only timestamp.

Response: unchanged. Both APIs return 55,193 rows for subnet 1, with the same values.

Replaced by: GET /v1/miners/coldkey-summary

Parameters: days can be at most 36,500. limit returns a 400 error.

Response: unchanged.

Differences: With days=7, every total matches. total_balance and the staked balances differ slightly (2903214897 against 2903227822 for the coldkey checked).

Replaced by: GET /v1/miners/weights

Parameters: mechanism is new. order_by accepts validator_uid (the default), netuid and miner_uid, and the default direction is ascending.

Response: mechanism is new. weight carries fewer digits (0.0026771653543307085 against 0.00267716535433070866). Rows with the same sort value may come back in a different order.

Replaced by: GET /v1/miners/weights/history

Parameters: mechanism is new. order_by accepts only timestamp.

Response: As on the latest route.

Differences: The new API keeps older history. For the pair checked it returns 1,221 rows going back to 2026-01-12; the old API returns 1,216 rows going back to 2026-09-12.

Replaced by: GET /v1/subnets/conviction

Parameters: order_by accepts amount_locked (the default), amount_tao, conviction and netuid, and the default direction is descending. Sorting by coldkey or hotkey has been removed.

Response: Rows are updated hourly. For a few seconds after an update a response can mix rows from two updates; each row's block_number shows which one it came from.

Differences: Both APIs return 395 rows with the same values.

Replaced by: GET /v1/subnets/conviction/history

Parameters: order_by accepts only timestamp. Sorting by coldkey or block_number has been removed.

Response: unchanged.

Differences: For subnet 79 both APIs return 260 rows, with the same values at block 9,227,386.

Replaced by: GET /v1/alpha/leaderboard

Parameters: coldkey, hotkey, netuid are unchanged. balance_min, balance_max, balance_as_tao_min and balance_as_tao_max are now whole numbers in RAO. Sorting is by global_rank (the default, largest position by TAO value first) or subnet_rank only; the old netuid, balance and balance_as_tao sorts are gone.

Response: New fields: global_rank (rank across all subnets by TAO value), locked_alpha and free_alpha (the part of balance held as miner registration collateral, and the part you can withdraw; they add up to balance).

Differences: The new API refreshes these positions once an hour, so a row can be up to an hour old; the old API was close to live. On the same position, balance was identical (665,012,418,306,538).

Replaced by: GET /v1/alpha/history

Parameters: coldkey, hotkey and netuid are still required. block_number is removed: use block_start/block_end or timestamp_start/timestamp_end. Sorting is by timestamp only.

Response: New fields: global_rank, subnet_rank, subnet_total_holders, locked_alpha, free_alpha.

Differences: The new API returns one row per day, the position at the day's last block (stamped 23:59:48). The old API also returned a row each time the position changed during the day. Over 26 to 29 Sep 2026 on one position, the old API returned 14 rows and the new one 4; the 4 end-of-day rows have the same values on both. You can no longer ask for the balance at an exact block.

Replaced by: GET /v1/alpha/portfolio

Parameters: Only coldkey (required), hotkey, netuid and days remain. balance_min, balance_max, balance_as_tao_min, balance_as_tao_max, page, limit and order are removed, and sending any of them is a 400.

Response: No pagination: every position for the coldkey comes back in one response. New fields: block_number, timestamp.

Differences: Rows can come back in a different order, and there is no way to choose it.

Partly replaced by: GET /v1/accounts/leaderboard and GET /v1/accounts/{address}

Parameters: To rank coldkeys by total stake, call /v1/accounts/leaderboard?order_by=balance_staked&order_dir=desc. total_balance_as_tao_min/_max become balance_staked_min/_max, whole numbers in RAO. To look up one coldkey, call /v1/accounts/{address} instead of passing coldkey.

Response: total_balance_as_tao is balance_staked (root stake plus alpha stake valued in TAO). On the same coldkey the two agreed: 282,655,058,255,990 against 282,655,126,926,410, a few blocks apart. The account endpoints return many more fields per account.

Differences: The new rank ranks accounts by total balance (free plus staked), not by stake, so it can differ from the old rank. The leaderboard lists every account, including ones with no stake; add balance_staked_min=1 to leave those out.

Replaced by: GET /v1/alpha/hotkey-shares

Parameters: Only netuid, hotkey, alpha_min (a whole number in RAO), page and limit remain. alpha_max and order are removed. Rows are always sorted by alpha, largest first.

Differences: The new API returns only hotkeys that hold shares on the subnet now, refreshed once an hour. The old API kept a hotkey's last row for ever after it left, so it returned far more rows, many of them out of date: on subnet 64, the old API had 2,797 rows and the new one 308, and the old API's largest row was last written in May 2026. Old rows written before about block 7,742,011 can show shares 2^64 times too large; the new API does not have this fault.

No replacement. The new API serves only each hotkey's current alpha and shares, through /v1/alpha/hotkey-shares, not their history.

Partly replaced by: GET /v1/alpha/leaderboard

Parameters: coldkey, hotkey and netuid are unchanged. alpha_min/alpha_max become balance_min/balance_max, whole numbers in RAO. Sorting is by global_rank or subnet_rank only.

Response: The old alpha is the new balance: on one position at block 9,227,203 both were 663,774,741,729,695. The old shares field is not served anywhere.

Partly replaced by: GET /v1/alpha/history

Parameters: coldkey, hotkey and netuid are all required. block_number is removed. Sorting is by timestamp only.

Response: The old alpha is the new balance. shares is not served.

Differences: One row per day, at the day's last block, instead of a row each time the position changed.

Replaced by: GET /v1/subnets/trades

Parameters: Every filter is kept. tao_value_min and tao_value_max are whole numbers in RAO. Sorting is by timestamp, from_amount, to_amount, tao_value or usd_value; the block_number sort is gone.

Response: New field: id. usd_value can be null.

Differences: On a buy, tao_value is now the TAO actually spent (for example 2,000,000,000), where the old API gave the value of the alpha received at the spot price (2,002,206,733). usd_value follows it. On a sell where part of the alpha paid the transaction fee, the new from_amount leaves the fee out: on extrinsic 9232001-0010 the old API said 1,105,235,540 and the new one 1,078,902,450, and to_amount moved with it. The new API also lists trades made through the EVM staking precompile, which the old API left out.

Partly replaced by: GET /v1/subnets/burns

Parameters: amount_min and amount_max are removed, and so are the netuid and amount sorts. burn_type accepts only call.

Differences: The new API lists only alpha burned by a call. The old API also served incentive burns (burn_type=incentive, 922,074 rows); asking the new API for them is a 400. Call burns are identical: 19,346 rows on subnet 64 on both, same values.

Replaced by: GET /v1/subnets/burns/total

Parameters: The amount sort is removed.

Differences: The new amount is the running total the chain itself keeps, so it can go down as well as up, and it starts again from zero when a new subnet takes over the netuid. The old total was counted another way and does not match: on subnet 64 the old API said 261,844,763,095,323 and the new one 274,207,517,615,115. The new API also lists subnet 0, with an amount of 0.

Replaced by: GET /v1/subnets/epochs

Parameters: block_number is removed: use block_start and block_end set to the same block. Sorting is by timestamp only.

Response: tao_in_pool is now tao_in_emission, alpha_in_pool is alpha_in_emission, and alpha_rewards is alpha_out_emission; the values matched on subnet 64 at block 9,231,883. name and symbol are removed. New fields: server_emission, validator_emission, root_alpha_divs and owner_cut, the amounts building up until the subnet's next payout.

Differences: The new API has a row for every block. The old API had one row every 360 blocks or so.

No replacement. No new endpoint gives emission per hotkey per subnet over time. The emission field on /v1/subnets/metagraph/history measures something else and does not match.

No replacement. The new API does not serve a subnet's TAO flow.

No replacement. The new API has no delegation volume endpoint. The old one currently answers 500 to every request.

No replacement. The new API has no slippage quote calculator. Each stake event on /v1/subnets/stake-events reports the slippage that trade actually had.

These three keep the TradingView UDF formats they had before. They do not use the new API's usual envelope, and they ignore parameters they do not know instead of answering 400.

Replaced by: GET /v1/tradingview/udf/config

Parameters: unchanged

Response: unchanged (identical byte for byte).

Replaced by: GET /v1/tradingview/udf/symbol_info

Parameters: unchanged

Response: unchanged (identical for one subnet and for all 130 symbols).

Replaced by: GET /v1/tradingview/udf/history

Parameters: unchanged (symbol such as SUB-64 or SUB--1, resolution, from, to, countback).

Differences: Each bar's time t is the start of its period; the old API labelled each bar one period later. Prices are given in full; the old API cut them to 6 decimal places. The bar that starts exactly at from is included. Volume v differs by a few percent in either direction. Weekly bars (7D) are 7-day periods starting on a Thursday and monthly bars (30D) are fixed 30-day periods, not calendar weeks and months. Up to 5,000 bars are returned.

Replaced by: GET /v1/price

Parameters: unchanged. asset ignores case.

Response: unchanged

Replaced by: GET /v1/price/simple

Parameters: unchanged

Response: unchanged

Replaced by: GET /v1/price/history

Parameters: Sorting is by timestamp only.

Response: unchanged

Replaced by: GET /v1/price/ohlc

Parameters: period is still required (1m, 1h, 1d).

Response: unchanged

Differences: timestamp_end is exclusive: a candle that starts exactly at timestamp_end is left out. The old API documented it as inclusive.

Replaced by: GET /v1/network/stats

Response: Returns a single object. Removed: subnet_registration_cost, staked_root_on_delegate, staked_root_on_keep, staked_root_on_partial_keep and staked_root_on_swap.

Differences: staked_alpha is about 24% lower than the old API's (1,984,028,268,463,932 against 2,596,919,576,010,487, 41 blocks apart). On the new API staked always equals staked_alpha plus staked_root; on the old API it did not. The other figures agree within a few blocks.

Replaced by: GET /v1/network/stats/history

Parameters: Sorting is by timestamp only.

Response: The same five fields as on /v1/network/stats are removed.

Differences: One row per day at the day's last block, as before, but the old API also returned a row for today so far; the new API's newest row is yesterday's. On recent days staked_alpha is about 24% lower than the old API's. Older rows can differ in accounts too: on 29 Sep 2025 the old API said 399,271 and the new one 352,371.

Replaced by: GET /v1/network/parameters

Response: Returns a single object. All 22 values are identical; only the order of the keys changed.

These four keep the CoinGecko DEX formats they had before. They do not use the new API's usual envelope, and they ignore parameters they do not know instead of answering 400.

Replaced by: GET /v1/coingecko/latest-block

Parameters: unchanged

Response: unchanged

Replaced by: GET /v1/coingecko/asset

Parameters: unchanged

Response: unchanged

Differences: A subnet token's totalSupply and circulatingSupply can be slightly lower: for subnet 64, 6,105,560.49 against the old API's 6,117,922.11, 14 blocks apart. Subnet 0 was identical.

Replaced by: GET /v1/coingecko/pair

Parameters: unchanged

Response: unchanged

Replaced by: GET /v1/coingecko/events

Parameters: unchanged. toBlock is still at most 20 blocks after fromBlock.

Response: unchanged

Differences: Swaps made through EVM transactions are included: blocks 9,232,000 to 9,232,010 gave 47 events against the old API's 45. priceNative can differ from about the seventh significant digit.

No replacement. The new API does not serve subnet developer activity.

No replacement. The new API does not serve subnet developer activity.

No replacement. The new API does not serve repository changelogs.

The old API's /v1 OTC routes and its /v2 OTC routes read two different OTC contracts, and they hold different data. The new API keeps them apart: the version-1 contract is under /v1/otc/contract-v1/..., and the version-2 contract is under /v1/otc/.... A version-1 row has an absolute price. A version-2 row has price_offset_bps instead, plus executed_price on fills.

Replaced by: GET /v1/otc/contract-v1/listings

Parameters: order → order_by + order_dir (columns: created_block, updated_block, price, amount; the old created_* and updated_* values are now created_block and updated_block). Everything else is unchanged, including price_min and price_max, which take RAO strings.

Response: seller and hotkey are plain SS58 strings. Timestamps always include milliseconds: 2025-12-03T17:06:36Z is now 2025-12-03T17:06:36.000Z. This is the same instant. price has the same digits as before (raw RAO per whole alpha, unscaled).

Differences: The seller and hotkey filters work now. On the old API they matched nothing. For example, seller=5E5Ctr2D9SjvLwNn45UNhBpjuQ7QWuinMqpAXY1ueRfJr5PT returns 0 rows on the old API and 1 row on the new one. Both SS58 and hex addresses are accepted.

Replaced by: GET /v1/otc/contract-v1/listings/history

Parameters: order → order_by + order_dir (columns: block_number, timestamp).

Response: seller, hotkey and buyer are plain SS58 strings. Timestamps always include milliseconds. On a cancelled row, amount and price are "0" and the returned alpha is in amount_returned. The old API returns these rows the same way.

Replaced by: GET /v1/otc/contract-v1/offers

Parameters: order → order_by + order_dir (columns: created_block, updated_block, price, amount; the old created_* and updated_* values are now created_block and updated_block).

Response: buyer is a plain SS58 string. Timestamps always include milliseconds.

Differences: The buyer filter now matches. On the old API it matched nothing, the same fault as /api/otc/listing/v1. This one comes from the new API's documentation and was not re-run against this route.

Replaced by: GET /v1/otc/contract-v1/offers/history

Parameters: order → order_by + order_dir (columns: block_number, timestamp).

Response: buyer and seller are plain SS58 strings. Timestamps always include milliseconds.

Replaced by: GET /v1/otc/contract-v1/trades

Parameters: order → order_by + order_dir (columns: block_number, timestamp, tao_amount, alpha_amount).

Response: seller and buyer are plain SS58 strings. Timestamps always include milliseconds. There is no executed_price, and the old version-1 route had none either.

Replaced by: GET /v1/otc/contract-v1/users/stats

Parameters: order → order_by + order_dir (columns: last_activity_block, volume_tao, volume_alpha, total_trades). last_activity_* is now last_activity_block.

Response: account is a plain SS58 string. Timestamps always include milliseconds.

Differences: The account filter works now. On the old API it matched nothing: account=5E5Ctr2D9SjvLwNn45UNhBpjuQ7QWuinMqpAXY1ueRfJr5PT returns 0 rows on the old API and 1 row on the new one. This route and /api/otc/user/stats/v2 hold different data (7 rows against 1 today), so do not swap one for the other.

Replaced by: GET /v1/otc/contract-v1/subnets/status

Parameters: order → order_by + order_dir (columns: netuid, timestamp, block_number). frozen still takes frozen, unfrozen or all. It is not a boolean.

Response: changed_by is a plain SS58 string. Timestamps always include milliseconds.

Differences: Both APIs return no rows today. The field list was compared from the two specifications, and the two lists are the same.

Replaced by: GET /v1/otc/listings

Parameters: order → order_by + order_dir (columns: created_block, updated_block, price_offset_bps, amount; the old created_* and updated_* values are now created_block and updated_block).

Response: seller and hotkey are plain SS58 strings. Timestamps always include milliseconds.

Differences: Both APIs return no rows today. The field list was compared from the two specifications, and the two lists are the same.

Replaced by: GET /v1/otc/listings/history

Parameters: order → order_by + order_dir (columns: block_number, timestamp).

Response: seller, hotkey and buyer are plain SS58 strings. Timestamps always include milliseconds.

Differences: Both APIs return no rows today. The field list was compared from the two specifications, and the two lists are the same.

Replaced by: GET /v1/otc/offers

Parameters: order → order_by + order_dir (columns: created_block, updated_block, price_offset_bps, amount; the old created_* and updated_* values are now created_block and updated_block).

Response: buyer is a plain SS58 string. Timestamps always include milliseconds.

Replaced by: GET /v1/otc/offers/history

Parameters: order → order_by + order_dir (columns: block_number, timestamp).

Response: buyer and seller are plain SS58 strings. Timestamps always include milliseconds.

Replaced by: GET /v1/otc/trades

Parameters: order → order_by + order_dir (columns: block_number, timestamp, tao_amount, alpha_amount, executed_price).

Response: seller and buyer are plain SS58 strings. Timestamps always include milliseconds.

Differences: Both APIs return no rows today. The field list was compared from the two specifications, and the two lists are the same.

Replaced by: GET /v1/otc/users/stats

Parameters: order → order_by + order_dir (columns: last_activity_block, volume_tao, volume_alpha, total_trades). last_activity_* is now last_activity_block.

Response: account is a plain SS58 string. Timestamps always include milliseconds.

Replaced by: GET /v1/otc/subnets/status

Parameters: order → order_by + order_dir (columns: netuid, timestamp, block_number). frozen still takes frozen, unfrozen or all.

Response: changed_by is a plain SS58 string. Timestamps always include milliseconds.

Differences: Both APIs return no rows today. The field list was compared from the two specifications, and the two lists are the same.

Replaced by: GET /v1/otc/lockup/listings

Parameters: order → order_by + order_dir (columns: created_block, updated_block, price_offset_bps, lockup_duration, total_amount, remaining_amount; the old created_* and updated_* values are now created_block and updated_block).

Response: seller and hotkey are plain SS58 strings. Timestamps always include milliseconds.

Replaced by: GET /v1/otc/lockup/listings/history

Parameters: order → order_by + order_dir (columns: block_number, timestamp).

Response: seller and buyer are plain SS58 strings. escrow_account was a hex public key (0xc407…c13d) and is now the SS58 address of the same key (5GVjWfjhfg6hPcLJYeJezExwE7ZmqLmvDKtYJ5EqAB5TF3ev). Timestamps always include milliseconds.

Replaced by: GET /v1/otc/lockup/purchases

Parameters: order → order_by + order_dir (columns: created_block, unlock_block, alpha_amount, tao_amount, executed_price; the old created_* value is now created_block).

Response: buyer and seller are plain SS58 strings. escrow_account was a hex public key and is now the SS58 address of the same key. Timestamps always include milliseconds.

Replaced by: GET /v1/otc/lockup/claims

Parameters: order → order_by + order_dir (columns: block_number, timestamp, amount_claimed, unlock_block). escrow_account accepts an SS58 address or a hex key.

Response: buyer is a plain SS58 string. escrow_account is an SS58 address. On the old API's other lockup routes it was a hex key.

Differences: Neither API has a claim row today. The field list was compared from the two specifications, and the two lists are the same.

Replaced by: GET /v1/otc/lockup/users/stats

Parameters: order → order_by + order_dir (columns: last_activity_block, total_listings_created, total_purchases_made, total_claims_made, volume_sold_tao, volume_bought_tao). Three sort columns are renamed: last_activity_* is now last_activity_block, total_volume_sold_tao_* is now volume_sold_tao, and total_volume_bought_tao_* is now volume_bought_tao. The response fields keep their total_volume_… names.

Response: account is a plain SS58 string. Timestamps always include milliseconds.

On the old API, the /api/v1/live/... routes read straight from a chain node and returned decoded JSON. The new API has no matching REST routes. Some of the data is in its indexed routes (/v1/blocks, /v1/extrinsics, /v1/events, /v1/accounts/{address}). Everything else has to be read from the node yourself, through the new API's JSON-RPC pass-through: POST /v1/rpc/http, or the WebSocket wss://…/v1/rpc/ws/{target} with target finney_lite or finney_archive. The HTTP route always uses finney_lite. To read state at older blocks, use the WebSocket with finney_archive. The pass-through returns the node's own JSON-RPC answer, so storage values and metadata come back SCALE-encoded and you have to decode them. We did not call the pass-through when checking this guide, because it is a POST.

Partly replaced by: GET /v1/blocks?limit=1. Rows come newest first by default.

Parameters: unchanged (none).

Response: A paginated list with one row, not a bare object. Renamed: number (string) → block_number (number), parentHash → parent_hash, stateRoot → state_root. extrinsicRoot was always null on the old route; the new extrinsics_root is filled in. authorId was null; the new validator is also null. Removed: finalized, logs, onInitialize, onFinalize, extrinsics. Added: timestamp, spec_version, spec_name, impl_name, impl_version, events_count, extrinsics_count, calls_count.

Differences: The new row does not include the block's extrinsics and events. Get them from GET /v1/extrinsics?block_number=<n>&include_args=true and GET /v1/events?block_number=<n>. The new API serves the newest block it has indexed, not the node's head. In our check the two were the same block, 9,232,197.

Partly replaced by: GET /v1/blocks?block_number={height}

Parameters: the path height → query block_number.

Response: The same changes as /api/v1/live/blocks/head. On block 9,232,197, hash, parent_hash and state_root match the old values. The 266 events and 34 extrinsics in the old block body match the new events_count (266) and extrinsics_count (34).

Differences: The new API indexes blocks from genesis.

Partly replaced by: GET /v1/blocks?block_start=<a>&block_end=<b>

Parameters: block_start and block_end are unchanged. The new route also takes page, limit, order_by and order_dir, plus timestamp_start, timestamp_end, block_number and hash.

Response: A paginated data list, not a bare array. Each row changes as described for /api/v1/live/blocks/head.

Differences: The new API returns at most 200 blocks a page.

Partly replaced by: GET /v1/extrinsics?id={height}-{index}&include_args=true. The index is zero-padded to four digits, for example id=9232197-0018. For that extrinsic's events, use GET /v1/events?extrinsic_id={height}-{index}.

Parameters: The two path values become one id query value.

Response: Pallet and call names are now in runtime spelling, so subtensorModule / addStakeLimit is now pallet: "SubtensorModule", name: "add_stake_limit". Numbers inside args are JSON numbers, where the old API used strings ("30000000" is now 30000000). Added: id, index, block_number, timestamp, signer_address, fee, fee_payer, error, args_summary. Events are not embedded; they are on /v1/events, with named args instead of a positional data array.

Differences: The old route returned HTTP 500 on all three calls we made, on two different blocks. We matched values against the same extrinsic as it appears inside the old /api/v1/live/blocks/{height} body: hash, args and signer all match on extrinsic 18 of block 9,232,197.

No replacement. The new REST routes do not carry raw encoded extrinsics. The node's chain_getBlock method, called through POST /v1/rpc/http, returns the header and the hex extrinsics (not checked live). The header roots are on GET /v1/blocks: on block 9,232,197 the old extrinsicRoot matches the new extrinsics_root.

Partly replaced by: GET /v1/accounts/{address}. It returns a list with one row.

Parameters: unchanged (address in the path).

Response: free → balance_free and reserved → balance_reserved. These are the latest indexed values, not a live node read. Removed: nonce, frozen, miscFrozen, feeFrozen, locks, tokenSymbol, at. No new REST route carries those. To get them, read System.Account through the JSON-RPC pass-through (not checked live). The new row adds staked balances, alpha positions and 24-hour-ago values.

Differences: The old route returned HTTP 500 for both addresses we tried, so we could not compare values.

No replacement. No new REST route returns pending transactions. The node method author_pendingExtrinsics can be called through POST /v1/rpc/http, if the node allows it (not checked live).

No replacement. The node's client name and version (chain, clientImplName, clientVersion) are not on any new REST route. They can be read with the node methods system_chain, system_name and system_version through POST /v1/rpc/http (not checked live). GET /v1/network/runtime-version is a different thing: it returns the chain's runtime version number (runtime_version). Returns a single object.

No replacement. Pallet constants come from the chain metadata. Read it with state_getMetadata through POST /v1/rpc/http. It comes back SCALE-encoded, and you have to decode it yourself (not checked live).

No replacement. It is the same as /api/v1/live/pallets/{pallet_id}/consts: read it from state_getMetadata through the JSON-RPC pass-through.

Differences: The old route returned "metadata": null for balances / ExistentialDeposit under both spellings we tried, so it was not returning the constant.

No replacement. Event definitions (names, fields, docs) come from the chain metadata, through state_getMetadata on the JSON-RPC pass-through. To find events that actually happened, use GET /v1/events?pallet=<Pallet>&name=<Event>. That is a different dataset.

No replacement. It is the same as /api/v1/live/pallets/{pallet_id}/events.

Differences: The old route returned "metadata": null for balances / Transfer under both spellings we tried.

No replacement. Storage item definitions come from the chain metadata, through state_getMetadata on the JSON-RPC pass-through.

No replacement. No new REST route reads an arbitrary storage item. Call state_getStorage with the item's storage key through POST /v1/rpc/http. The value comes back SCALE-encoded hex, not decoded JSON as on the old route (not checked live). Many common storage values are already served decoded by the new API's indexed routes. For example, total issuance is on /v1/network/stats.

These are new. Nothing in the old API maps to them.

Cross-subnet alpha swaps: one row per swap, with coldkey, from_name and to_name (for example SN41 and SN51), from_amount, to_amount, tao_value (RAO, as strings) and usd_value. Filter by coldkey, extrinsic_id, from_name, to_name, value ranges (tao_value_min/_max, usd_value_min/_max), and block or time range. Paged as usual.

One bar per UTC day of alpha burned on one subnet, built from the chain's running burn total. Parameters: netuid and days.

These follow CoinMarketCap's own formats: no data/pagination wrapper.

• GET /v1/cmc/assets: every tradeable subnet token, keyed by symbol (SN0, SN1, ...).
• GET /v1/cmc/summary: a market summary for every active trading pair (trading_pairs, last_price, lowest_ask and so on), as a list.
• GET /v1/cmc/ticker: a ticker for every active trading pair, keyed by id.
• GET /v1/cmc/circulating-supply?netuid=<n> and GET /v1/cmc/total-supply?netuid=<n>: one bare JSON number.
• GET /v1/cmc/trades/market-pair?market_pair=<pair>: trades on one pair in the last five minutes. The pair is written as it appears in trading_pairs, for example TAO_SN0; any other form is a 400 Invalid market_pair format.
• GET /v1/cmc/order-book/market-pair?market_pair=<pair>: always an empty list. There is no order book.