Head to head · Exchange market data · October 2026 research run

Bybit vs Hyperliquid

Bybit scores 57.3 (C) on agent readiness against Hyperliquid's 50 (D), and leads in 4 of 7 scored categories. Hyperliquid leads on payments & pricing. Both do exchange market data.

Best crypto exchange APIs and MCP servers for AI agents · All 210 exchanges comparisons

Which one, for what

Bybit C

Best for An agent trading spot, perpetuals, futures and options in a person's Bybit account outside the excluded jurisdictions, with a demo environment to rehearse on.

Ahead on

  • Reliability, 45 against 40
  • Schema & documentation, 73 against 42
  • Maintenance & community, 66 against 39
  • Transparency & trust, 51 against 26

Also in its favour

  • Runs on your own machine

Watch for

No status page, incident history or SLA was found; the terms say the API is not guaranteed to be available

Hyperliquid D

Best for An agent trading perpetuals or spot on chain from its own wallet with no account, or acting for a master account through an approved API wallet.

Ahead on

  • Payments & pricing, 55 against 40

Watch for

No official MCP server; the 23 registry entries are all third parties, several selling data by x402

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Schema & documentation Bybit +31

Bybit 73

No OpenAPI or other machine-readable contract for the REST API; the MCP server's 398 tools have zod schemas turned into JSON Schema, so the contract exists for the MCP surface only (12 of 25). No llms.txt; the docs are a public Docusaurus repository (bybit-exchange/docs) whose MDX source an agent can read (5 of 10). The MCP tool descriptions state purpose, rules and an agent hint per tool, and the API reference gives one page per endpoint (15 of 20). Parameters are typed with enums such as category and orderType and required flags (12 of 15). Request and response examples on each endpoint page and an error code page of about 750 codes with text (14 of 15). A dated changelog with 24 entries in the last 90 days, versioned V3 to V5 (15). 73.

Hyperliquid 42

No OpenAPI or other machine-readable contract; request and response shapes are tables and JSON examples in the docs (0 of 25). llms.txt and a Markdown twin of every page by appending .md (10). Each info type and exchange action has a short description of what it does, and the signing page lists common mistakes, but little says when to use one call over another (12 of 20). Fields are one-letter keys (a, b, p, s, r, t, c) with numbers as strings, enums such as tif Alo, Ioc or Gtc, and required fields marked in tables (9 of 15). An example for every call and a list of order and cancel error strings, with no HTTP error table (11 of 15). No API version, no changelog and no dated release notes; deprecated actions such as userDexAbstraction are marked in place (0 of 15). Total 42.

Maintenance & community Bybit +27

Bybit 66

Read as a hosted service with an open-source MCP server. The MCP package shipped 2.1.26 on 30 October 2026 and the V5 changelog's latest entry is 29 October (30). 31 npm versions since 16 April 2026, 11 of them in the last 90 days, and 24 changelog entries in the same window (20). The trading-mcp repository has 4 open issues and 18 forks; whether issues get answers was not checked, and the API has no public forum we could read (10 of 25). Not in the official MCP registry (0). The package has four dependencies, no install scripts, a publish workflow and no test workflow (6 of 10). 66.

Hyperliquid 39

Read as a closed hosted service with open SDKs. The API has no dated changelog, so the last dated change is the Python SDK's 0.24.0 on 4 June 2026, 152 days before this check; the node repository was pushed on 10 August 2026 and the docs describe newer actions (HIP-4 outcome markets, borrow and lend) without dates. Judgement call (10 of 30). No three dated releases in 90 days (0). Support through Discord and the docs' FAQ, 112 open issues on the SDK whose replies couldn't be read (8 of 25). Official Python and Rust SDKs, the Rust one last pushed 21 October 2025; no MCP registry entry (15). CI on Python 3.9 and 3.13 with Poetry, Dependabot badge, a lock file (6 of 10). Total 39.

Transparency & trust Bybit +25

Bybit 51

The editorial half. A closed hosted service. The terms read are the UAE entity's PDF served from bybit.com, clear on API use and availability, but the global terms page renders by script and was not read; the MCP server is MIT (12 of 30). The privacy policy was not read, so data handling is unscored (5 of 30). The changelog dates removals and response changes but gives no notice period or deprecation policy (10 of 20). No subprocessor list or data location statement was read (0 of 20). 27.

Hyperliquid 26

The editorial half. The SDKs are MIT and the node repository Apache-2.0 with a Dockerfile around a binary; the API servers and front end are closed and the terms were unreadable (12 of 30). No privacy policy read and nothing in the docs on data handling (0 of 30). Deprecated actions are marked in the docs without dates or a policy (8 of 20). No subprocessors or data locations (0 of 20). Total 20.

Payments & pricing Hyperliquid +15

Bybit 40

No machine payment protocol (0). The fee schedule is public without login, with VIP 0 maker and taker rates per product and the tier criteria (20). Reading market data needs no key and the demo environment costs nothing, no card is asked for at sign-up (20). A person must register, pass identity verification and mint the key in a browser (0). 40.

Hyperliquid 55

No machine payment protocol (0). Fees published per tier without login (20). No card anywhere; public info endpoints are free and the testnet faucet gives 1,000 mock USDC, though only to an address that has deposited on mainnet (15 of 20). No signup; an agent with a funded EVM wallet deposits USDC on a bridged chain and trades, subject to the unread terms (20). Total 55.

Reliability Bybit +5

Bybit 45

Graded with the hosted lines on the V5 API at api.bybit.com. No status page found (status.bybit.com does not resolve, the docs link none), so 0 for the page and 5 for no readable history. Rate limits are published per endpoint and per UID with numbers, plus a 600 per 5 second IP limit (15). Every response carries X-Bapi-Limit, X-Bapi-Limit-Status and X-Bapi-Limit-Reset-Timestamp, retCode 10006 and HTTP 403 are documented with a wait rule, and orders take a unique orderLinkId that amend and cancel accept (15). No SLA (0). V5 is the current production API, not a beta (10). 45.

Hyperliquid 40

Read with the hosted lines on the REST API at api.hyperliquid.xyz, the surface an agent calls; the chain behind it is noted under details. No status page found in the docs, on hyperliquid.xyz or at status.hyperliquid.xyz, which did not resolve (0). No readable history (5). Rate limits documented with numbers per IP, per address and per WebSocket connection (15). When limited, an address gets one request every 10 seconds; no 429 shape or Retry-After is documented, but cloid, per-signer nonces, expiresAfter and noop make retries safe (10 of 15). No SLA (0). Mainnet is generally available (10). 0 + 5 + 15 + 10 + 0 + 10 = 40.

Score by category

CategoryWeight this runBybitHyperliquidEdge
Reliability16%204540Bybit +5
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%16.27342Bybit +31
Agent ergonomics13%16.26670Hyperliquid +4
Security & auth14%17.56064Hyperliquid +4
Payments & pricing10%12.54055Hyperliquid +15
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8.86639Bybit +27
Transparency & trust7%8.85126Bybit +25
Negative events≤1500
TotalC 57.3/100D 50/100

Facts side by side

FactBybitHyperliquid
KindHTTP APIHTTP API
VendorBybitHyperliquid Labs
Hosted endpointno (local only)no (local only)
Transportsstdio
AuthAPI keyOAuth or key
PricingPay per usePay per use
x402nono
LicencenoneThe exchange runs on the Hyperliquid chain (HyperCore), whose node is distributed as a binary under an Apache-2.0 repository; the Python and Rust SDKs are MIT. The API servers and front end are closed
Tools exposed3980
Read-only variant documentedyesno
llms.txtnoyes
Last release2026-10-302026-06-04
Terms last updatedcouldn't be readcouldn't be read
Privacy policy last updatedcouldn't be readno document linked
Customer content may train modelscouldn't be readcouldn't be read
Terms restrict automated accesscouldn't be readcouldn't be read
Terms restrict benchmarkingcouldn't be readcouldn't be read
Terms or service can change without noticecouldn't be readcouldn't be read
Arbitration or class-action waivercouldn't be readcouldn't be read
Popularity43 stars1.9k stars, 167k PyPI/wk

Verdicts

Bybit

Bybit publishes per-endpoint rate limits, a demo environment, scoped keys with a read-only flag and IP binding, and an official stdio MCP server whose 398 tools carry read-only and destructive annotations. There is no status page, no OpenAPI file, no SLA, the MCP server has no toolsets, and the global terms and privacy pages are drawn by script.

Hyperliquid

Hyperliquid's API needs no signup or key, an approved API wallet can trade but not withdraw, fees and rate limits are published with numbers, and a testnet mirrors mainnet. There is no official MCP server, OpenAPI file, status page, changelog or readable terms, and signing uses two schemes the docs say to leave to an SDK.

Before you call either

Bybit

  1. Create a key with only the permission groups the task needs, set readOnly for research keys, leave Withdraw off and bind an IP, or the key expires after 90 days
  2. Send a unique orderLinkId on every order and reuse it when amending or cancelling, since create-order is asynchronous and confirmed over WebSocket
  3. Read X-Bapi-Limit-Status before bursts and wait until X-Bapi-Limit-Reset-Timestamp on retCode 10006; on HTTP 403 stop all sessions for at least 10 minutes
  4. Test on api-demo.bybit.com with a demo key from the production account, not on testnet demo mode, which the docs call meaningless
  5. Run the MCP server with BYBIT_TESTNET=true first, and only in a client that shows a per-call approval, because the server has no read-only mode of its own

Hyperliquid

  1. Generate a fresh API wallet, approve it with approveAgent with a name and expiry, sign with it, and query with the master address, never the agent address
  2. Use the Python SDK's sign_l1_action and sign_user_signed_action; mixing the two, field order in msgpack or an uppercase address gives "User or API Wallet ... does not exist"
  3. Set a cloid on every order and read the per-order statuses array; one batch-level error rejects the whole batch
  4. Budget address limits by volume (one action per 1 USDC traded after 10,000); batch orders to lower IP weight, and use WebSocket for data
  5. Arm scheduleCancel before unattended trading so open orders are cancelled if the agent stops

Questions

Which is better for AI agents, Bybit or Hyperliquid?

Bybit scores 57.3 (C) on agent readiness against Hyperliquid's 50 (D), and leads in 4 of 7 scored categories. Hyperliquid leads on payments & pricing.

Do Bybit and Hyperliquid need an API key?

Bybit needs an API key. Hyperliquid takes an API key or an OAuth sign-in.

Can an agent call Bybit and Hyperliquid without installing anything?

Bybit runs on your own machine, with no hosted endpoint listed. No hosted endpoint is listed for Hyperliquid.

Other comparisons with Bybit or Hyperliquid

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.