Model APIs - 42 entries
Anthropic API rate limits
Anthropic's official docs, read on October 17, 2026, set rate limits for the Anthropic API across the Start, Build, Scale, Custom and Evaluation tiers. For Claude Fable 5.x (Messages API), Start tier is listed at 1,000 RPM, 500,000 ITPM and 100,000 OTPM, and Scale tier at 4,000 RPM, 4,000,000 ITPM and 800,000 OTPM.
Last verified from platform.claude.com. What the docs leave out
- Tiers
- 5
- Tables
- 17
- Entries
- 42
- Verified
- Oct 17, 2026
Limits by model, endpoint and plan
Values are copied exactly as the docs show them, units included, in the docs' own order. A dash means the docs list no value for that row.
All 17 tables on this page
- Claude Fable 5.x (Messages API)
- Claude Opus 5.5 (Messages API)
- Claude Opus 5 (Messages API)
- Claude Opus 4.x (Opus 4.8, 4.7, 4.6, 4.5 combined) (Messages API)
- Claude Sonnet 5.5 (Messages API)
- Claude Sonnet 5 (Messages API)
- Claude Sonnet 4.x (Sonnet 4.6 and 4.5 combined) (Messages API)
- Claude Haiku 5.5 (Messages API)
- Claude Haiku 4.5 (Messages API)
- Claude Haiku 3.5 (retired, except on Google Cloud) (Messages API)
- Organization (all API usage)
- Message Batches API (shared across all models)
- Claude Managed Agents endpoints
- Files API (upload, list, retrieve, download, delete)
- Fast mode (speed: "fast") on Claude Opus 5.5, Claude Opus 5, Claude Opus 4.8
- All models (organization)
- Claude Platform on AWS (all models)
Claude Fable 5.x (Messages API)
3 tiers
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 2,000 |
| |
| Scale tier | 4,000 |
| |
- Fable rate limit is a total limit that applies to combined traffic across Claude Fable 5.1 and Claude Fable 5. Claude Mythos 5.1 and Claude Mythos 5 share a separate combined limit on the same terms. Only uncached input tokens (input_tokens + cache_creation_input_tokens) count toward ITPM. Org-level, token bucket algorithm, shared across all inference_geo values. Exceeding returns HTTP 429.
Claude Opus 5.5 (Messages API)
3 tiersModel page: Claude Opus 5.5
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Separate rate limit, not part of the Opus 4.x combined bucket. Only uncached input tokens count toward ITPM. Fast mode has a separate dedicated limit.
Claude Opus 5 (Messages API)
3 tiersModel page: Claude Opus 5
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Separate rate limit, not part of the Opus 4.x combined bucket. Only uncached input tokens count toward ITPM.
Claude Opus 4.x (Opus 4.8, 4.7, 4.6, 4.5 combined) (Messages API)
3 tiers
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Opus rate limit is a total limit that applies to combined traffic across Claude Opus 4.8, Opus 4.7, Opus 4.6, and Opus 4.5. Claude Opus 5.5 and Claude Opus 5 each have a separate rate limit.
Claude Sonnet 5.5 (Messages API)
3 tiersModel page: Claude Sonnet 5.5
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Separate rate limit, not part of the Sonnet 4.x combined bucket. Only uncached input tokens count toward ITPM.
Claude Sonnet 5 (Messages API)
3 tiersModel page: Claude Sonnet 5
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Separate rate limit, not part of the Sonnet 4.x combined bucket.
Claude Sonnet 4.x (Sonnet 4.6 and 4.5 combined) (Messages API)
3 tiers
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Sonnet 4.x rate limit is a total limit that applies to combined traffic across Sonnet 4.6 and Sonnet 4.5 (deprecated).
Claude Haiku 5.5 (Messages API)
3 tiersModel page: Claude Haiku 5.5
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Only uncached input tokens count toward ITPM.
Claude Haiku 4.5 (Messages API)
3 tiersModel page: Claude Haiku 4.5
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 5,000 |
| |
| Scale tier | 10,000 |
| |
- Only uncached input tokens count toward ITPM.
Claude Haiku 3.5 (retired, except on Google Cloud) (Messages API)
3 tiers
| Tier | RPM |
|---|---|
| Start tier | 1,000 |
| |
| Build tier | 2,000 |
| |
| Scale tier | 4,000 |
| |
- Limit counts cache_read_input_tokens toward ITPM usage. Model retired, except on Google Cloud.
Organization (all API usage)
4 tiers
| Tier | Monthly spend cap |
|---|---|
| Start tier | $500 USD |
| Build tier | $1,000 USD |
| Scale tier | $200,000 USD |
| Custom tier | No monthly spend cap; limits are arranged with their account team |
| |
- Start tier, Build tier and Scale tier: At the cap, API usage pauses until 00:00 UTC on the first day of the next month; requests return HTTP 429 with error type rate_limit_error, error.details.error_code "enforced_spend_limit_reached", and no retry-after header. A user-set spend limit below the cap returns HTTP 400 invalid_request_error with a message beginning "You have reached your specified API usage limits".
- Custom tier: Applies to Messages API and Message Batches API limits above Scale.
Claude Managed Agents endpoints
| Metric | Limit |
|---|---|
| Create endpoints (for example, agents, sessions, and environments) | 300 requests per minute |
| Read endpoints (for example, retrieve, list, and stream) | 1,200 requests per minute |
- Rate-limited per organization; separate from Messages API rate limits. No per-tier variation published.
Files API (upload, list, retrieve, download, delete)
| Metric | Limit |
|---|---|
| RPM | approximately 500 requests per minute |
- Verbatim: "File-related API calls are limited to approximately 500 requests per minute." Per-organization, shared across file operations, separate from Messages API limits.
Fast mode (speed: "fast") on Claude Opus 5.5, Claude Opus 5, Claude Opus 4.8
Start, Build, ScaleModel pages: Claude Opus 5, Claude Opus 4.8
| Metric | Start, Build, Scale |
|---|---|
| Fast mode input / output tokens per minute | Dedicated limit separate from standard Opus rate limits; values shown only on the Rate limits page in the Claude Console |
- Claude Opus 5.5 has a default fast mode rate limit in the Start, Build, and Scale tiers (no access request needed); Evaluation tier organizations can use it once they move to Start, Build, or Scale. Fast mode for Claude Opus 5 and Claude Opus 4.8 is in research preview. Exceeding returns 429 with retry-after. Headers: anthropic-fast-input-tokens-limit, anthropic-fast-input-tokens-remaining, anthropic-fast-input-tokens-reset, anthropic-fast-output-tokens-limit, anthropic-fast-output-tokens-remaining, anthropic-fast-output-tokens-reset. Not available on Claude Platform on AWS.
All models (organization)
Evaluation tier
| Metric | Evaluation tier |
|---|---|
| RPM / ITPM / OTPM | Below the standard limits shown on the rate limits page; exact values not published |
- New organizations and organizations with limited usage history may start in the Evaluation tier as part of fraud and abuse prevention.
Claude Platform on AWS (all models)
Start tier (initial placement)
| Metric | Start tier (initial placement) |
|---|---|
| Rate limits and monthly spend caps | Same per-tier limits as the first-party Claude API |
- Per-workspace rate limit configuration and fast mode are not available on Claude Platform on AWS. It uses a separate capacity pool from the first-party Claude API and Amazon Bedrock.
How to raise your limits
Claude Fable 5.x (Messages API), Claude Opus 5.5 (Messages API) and 10 more
Tiers advance automatically based on usage history and account standing. To request higher rate limits or a higher monthly spend cap, use Request tier increase on the Rate limits page in the Claude Console (platform.claude.com/usage/limits); Anthropic support can also raise limits. Above Scale: contact sales via the Rate limits page (Custom tier).
Files API (upload, list, retrieve, download, delete)
To request a higher limit, contact sales (sales@anthropic.com).
All models (organization)
Limits increase automatically as the organization builds usage history.
Claude Platform on AWS (all models)
Organizations move to a higher tier automatically as they build a history of paid AWS Marketplace invoices. To request higher limits, contact your Anthropic account representative or Anthropic support, including the models needed, peak input and output tokens per minute per model, and roughly what share of input is cached or repeated context. The Request tier increase flow is not available.
What the docs leave out
- Organization-specific current tier and limits (including Custom tier values and fast mode token-per-minute values) are shown only in the logged-in Claude Console Rate limits page (platform.claude.com/usage/limits); public docs give only the standard per-tier tables listed here.
- Evaluation tier exact RPM/ITPM/OTPM values are not published (stated only as below the standard limits).
- No published RPD (requests per day) or concurrency limits for the Messages API; acceleration limits on sharp usage increases exist but no values are published.
- Tier advancement thresholds (spend or deposit amounts to move Start -> Build -> Scale) are not on the rate limits page; it states only that tiers advance automatically based on usage history and account standing.
- Claude Mythos 5.1 / Mythos 5 combined limit values are not tabulated separately; docs say they share a separate combined limit on the same terms as Fable 5.x.
FAQ
What are the Anthropic API rate limits?
Anthropic's official docs, read on October 17, 2026, set rate limits for the Anthropic API across the Start, Build, Scale, Custom and Evaluation tiers. For Claude Fable 5.x (Messages API), Start tier is listed at 1,000 RPM, 500,000 ITPM and 100,000 OTPM, and Scale tier at 4,000 RPM, 4,000,000 ITPM and 800,000 OTPM.
How do I raise my Anthropic API rate limits?
Tiers advance automatically based on usage history and account standing. To request higher rate limits or a higher monthly spend cap, use Request tier increase on the Rate limits page in the Claude Console (platform.claude.com/usage/limits); Anthropic support can also raise limits. Above Scale: contact sales via the Rate limits page (Custom tier). To request a higher limit, contact sales (sales@anthropic.com). Limits increase automatically as the organization builds usage history.
Where do these Anthropic API numbers come from?
Every value is copied verbatim from Anthropic's official documentation at platform.claude.com, read on October 17, 2026. This page lists all 42 entries in 17 tables, each linked to the page it came from.
Related reading
- Claude Usage Limits With Fable 5: Windows and Burn RatesFable 5 drains Claude's 5-hour rolling usage window far faster than Opus or Sonnet. What the plan multipliers mean, the June 22 change, and how to make it last.
- Claude Fable 5 API: Production Integration Patterns, Rate Limits, and Migration GotchasClaude Fable 5 API production guide - adaptive thinking, rate limit strategy, streaming latency, fallback chains and a migration checklist from Opus 4.8.
- Claude Code Usage Limits in 2026: The Practical Playbook for Pro and Max TeamsClaude Code usage limits are mostly an ops problem. A playbook for Pro and Max teams: lock auth mode, bound subagents, add hooks and route work by difficulty.
- Claude Code Idle Compaction: What It Does and How to Turn It OffClaude Code 2.1.299 compacts sessions over 200K tokens after about 54 idle minutes. How the cache math works, the setting to turn it off, and when to keep it.
- Tool Calling Under the Hood: Claude Messages vs OpenAI ResponsesHow tool calling works on the wire: JSON schema tools, parallel calls, tool_choice, strict mode and streaming deltas in Claude Messages vs OpenAI Responses.
- Claude Code Projects: Coordinator, Threads, and LimitsClaude Code Projects is open to waitlisted Pro and Max users. How the coordinator and cloud threads work, what they cost in plan limits and where it breaks.
Sources
Last verified . Every value is copied verbatim from the official pages below, read on that date.
Newsletter
Know when limits change
One email per week. New model releases, pricing and limit changes, and what they mean for what you build.