Anthropic raises Claude API rate limits and consolidates tiers into Start, Build, and Scale
Claude Platform usage tiers are Start, Build, and Scale, and Sonnet and Haiku rate limits match Opus at every tier, including Opus 5.5, Sonnet 5.5, and Haiku 5.5. Tier placement is automatic, based on usage history rather than spend, but each tier still carries a monthly spend cap.
Claude API usage tiers are Start, Build, and Scale, and at each one the current Opus, Sonnet, and Haiku models, including Opus 5.5, Sonnet 5.5, and Haiku 5.5, share the same published limits. Anthropic's own release notes date the change to June 26, 2026 and describe it as Claude Sonnet and Claude Haiku rate limits "now match Claude Opus at every usage tier," with most organizations moving to a higher tier and none receiving lower limits than before. (Claude Platform release notes, Claude Platform docs)
Key Takeaways
- Standard tiers are now Start, Build, and Scale, replacing the older numbered usage tiers. A Custom tier is arranged with an account team above Scale.
- Opus 5.5, Opus 5, Sonnet 5.5, Sonnet 5, Haiku 5.5, and the 4.x buckets carry identical limits at each tier: 1,000 RPM on Start, 5,000 on Build, 10,000 on Scale.
- Fable 5.x is one shared bucket across Fable 5.1 and Fable 5, with lower limits; Mythos 5.1 and Mythos 5 share a separate one.
- Tier placement is automatic and based on usage history and account standing, not on what you spend.
- Each tier still carries a monthly spend cap: $500 on Start, $1,000 on Build, $200,000 on Scale.
- New organizations may start in an Evaluation tier below the published limits, which rises automatically as account history builds.
- For most models, cached input tokens do not count toward your input-token limit, which usually matters more than the tier number.
- Acceleration limits can return a 429 even when you are inside your published limits, if traffic ramps sharply.
The Three Standard Tiers, and What Each Is Worth
Rate limits are set per organization, per model, and measured in three ways at once: requests per minute (RPM), input tokens per minute (ITPM), and output tokens per minute (OTPM). Exceeding any one of them returns a 429 naming the limit you hit, plus a retry-after header. These are the standard published limits for the Messages API.
| Tier | Model | RPM | ITPM | OTPM |
|---|---|---|---|---|
| Start | Opus 5.5, Opus 5, Opus 4.x, Sonnet 5.5, Sonnet 5, Sonnet 4.x, Haiku 5.5, Haiku 4.5 | 1,000 | 2,000,000 | 400,000 |
| Start | Fable 5.x (Fable 5.1 and Fable 5 combined) | 1,000 | 500,000 | 100,000 |
| Build | Opus 5.5, Opus 5, Opus 4.x, Sonnet 5.5, Sonnet 5, Sonnet 4.x, Haiku 5.5, Haiku 4.5 | 5,000 | 5,000,000 | 1,000,000 |
| Build | Fable 5.x (Fable 5.1 and Fable 5 combined) | 2,000 | 1,500,000 | 300,000 |
| Scale | Opus 5.5, Opus 5, Opus 4.x, Sonnet 5.5, Sonnet 5, Sonnet 4.x, Haiku 5.5, Haiku 4.5 | 10,000 | 10,000,000 | 2,000,000 |
| Scale | Fable 5.x (Fable 5.1 and Fable 5 combined) | 4,000 | 4,000,000 | 800,000 |
Anthropic's tables still carry a row for Claude Haiku 3.5, marked "retired, except on Google Cloud," at much lower limits (1,000 RPM and 100,000 ITPM on Start). If you still call it anywhere, move that traffic to Haiku 5.5 or Haiku 4.5.
Above Scale is the Custom tier, arranged through sales from the Rate limits page in the Console. Below Start is the Evaluation tier: Anthropic's docs say new organizations and those with limited usage history "may start in the Evaluation tier, with limits below the standard limits shown on this page while account history is established," and that those starting limits increase automatically. If you are standing up a new organization, budget for that floor rather than for the Start row.
Three footnotes in the tables matter more than they look:
- Fable 5.x is a shared bucket. Anthropic's docs say the Fable limit "applies to combined traffic across Claude Fable 5.1 and Claude Fable 5," and that Mythos 5.1 and Mythos 5 share a separate combined limit. Moving from Fable 5 to Fable 5.1 does not give you a fresh limit.
- Opus 4.x is a shared bucket. The Opus 4.x limit applies to combined traffic across Opus 4.8, 4.7, 4.6, and 4.5. Opus 5.5 and Opus 5 each have their own separate limit and are not part of that bucket. (See our Opus 5.5 coverage.)
- Sonnet 4.x is a shared bucket across Sonnet 4.6 and the deprecated Sonnet 4.5, and Sonnet 5.5 and Sonnet 5 are each separate the same way. Migrating a workload to Claude Sonnet 5.5 therefore gives you a fresh limit rather than a share of an existing one. (See our Sonnet 5.5 launch coverage.)
Limits also apply separately per model, so you can run several models up to their respective limits at the same time. They are shared across inference_geo values, though: requests sent with inference_geo: "us" and inference_geo: "global" draw from the same pool.
Spend Caps Still Exist, and That Is Not a Contradiction
The original announcement's "no longer based on API spend" and the documentation's spend-cap table are both accurate, because they describe different things. Placement is no longer earned by spend: Anthropic's docs say organizations "are placed on a tier automatically based on usage history and account standing and can move to a higher tier over time as they use the API." Consumption is still capped: each standard tier carries a maximum monthly API spend.
| Tier | Monthly spend cap |
|---|---|
| Start | $500 |
| Build | $1,000 |
| Scale | $200,000 |
| Custom | None; arranged with the account team |
Once you reach your tier's cap, API usage pauses until the next calendar month unless you request a higher limit. While paused, requests return a 429 with error type rate_limit_error but no retry-after header; error.details.error_code is enforced_spend_limit_reached, which is how you tell it apart from a rate limit. You can also set your own spend limit below the cap on the Billing page in the Console, which is the mechanism most teams actually want for cost control.
Two Limits That Bite Before Your Tier Does
Most throttling questions are answered by two rules that have nothing to do with which tier you are on.
Cached reads are usually free of your input limit. For most Claude models, only uncached input tokens count toward ITPM. Tokens after your last cache breakpoint (input_tokens) and cache-writes (cache_creation_input_tokens) count; cache reads (cache_read_input_tokens) do not. Anthropic's worked example: with a 2,000,000 ITPM limit and an 80% cache hit rate, you can effectively process about 10,000,000 total input tokens per minute. The exception is Claude Haiku 3.5, now retired except on Google Cloud, which does count cache reads toward ITPM. Prompt caching is therefore a larger throughput lever than a tier upgrade for most repeated-context workloads.
Acceleration limits are separate from rate limits. Anthropic's docs warn that you may get 429s "because of acceleration limits on the API if your organization has a sharp increase in usage," and advise ramping traffic gradually and keeping usage patterns consistent. A load test that jumps straight to your ceiling can fail while a steady workload at the same rate succeeds. Limits are enforced with a token bucket, so capacity replenishes continuously rather than resetting on a clock, and a 60 RPM limit may be enforced as roughly one request per second.
One smaller rule worth knowing: max_tokens does not factor into OTPM. Output limits are evaluated in real time against tokens actually generated, so setting a generous max_tokens carries no rate limit penalty.
Batches, Managed Agents, and Fast Mode Have Their Own Limits
The Messages API tables are not the whole picture. Three surfaces are limited separately.
| Surface | Start | Build | Scale |
|---|---|---|---|
| Message Batches RPM | 1,000 | 2,000 | 4,000 |
| Batch requests in processing queue | 200,000 | 300,000 | 500,000 |
| Batch requests per batch | 100,000 | 100,000 | 100,000 |
Managed Agents endpoints are limited per organization and independently of the Messages API: 300 requests per minute for create operations (agents, sessions, environments) and 1,200 for read operations (retrieve, list, stream). Fast mode (research preview) with speed: "fast" on Opus 5.5, Opus 5, or Opus 4.8 draws on dedicated limits separate from standard Opus limits, and reports status in anthropic-fast-* response headers.
Anthropic also notes that Claude Platform on AWS behaves differently: those organizations are placed on the Start tier and move up automatically only as they build a history of paid AWS Marketplace invoices, the self-service Request tier increase flow is unavailable, and per-workspace rate limit configuration and fast mode are not offered.
How to Read Your Own Limits
The published tables are standard limits, not your limits. Anthropic states plainly that all published limits "represent maximum allowed usage, not guaranteed minimums." Three ways to see the real numbers:
- The Console. The Rate limits page shows your organization's tier and current limits; the Usage page carries input-token and output-token rate limit charts, including your cache hit rate.
- The Rate Limits API, which returns the configured limits for your organization and workspaces programmatically.
- Response headers on every call.
anthropic-ratelimit-requests-limit,-remaining, and-resetcover requests;anthropic-ratelimit-input-tokens-*and-output-tokens-*cover throughput;anthropic-ratelimit-tokens-*reports whichever limit is currently most restrictive, including a Workspace limit if one applies.
To request more, use Request tier increase on the Rate limits page, which covers both rate limits and the monthly spend cap.
Practical Takeaways for Developers
- Check your tier before you plan capacity. Start, Build, and Scale differ by 10x on RPM and 5x on throughput, and a new organization may sit below Start entirely in the Evaluation tier.
- Cache before you upgrade. For most models an 80% cache hit rate multiplies effective input throughput about fivefold at the same tier, which is the cheaper lever.
- Ramp gradually. Acceleration limits punish step changes in traffic independently of your tier.
- Design for 429s regardless. Every rate limit returns a 429 with
retry-after, so backoff remains required at every tier. The monthly spend-cap 429 is the exception: it carries noretry-afterand reportserror_code: "enforced_spend_limit_reached"inerror.details, so retries, including the SDKs' automatic retries, fail until the cap resets. - Split workloads across models deliberately. Limits are per model: Opus 5.5, Opus 5, Sonnet 5.5, Sonnet 5, and Haiku 5.5 each sit outside the shared 4.x buckets, while Fable 5.1 and Fable 5 share one.
- Set a workspace limit if you share an organization. Workspace-level spend and rate limits protect other workspaces from one project's overuse, though they cannot be set on the default workspace and never exceed the organization limit.
Read next
Related coverage- Claude
Claude API Request Limit: 128 MB for Messages and Token Counting
Anthropic's Claude Platform docs list a 128 MB maximum request size for Messages and Token Counting on the first-party API. Cloud-platform limits are smaller and vary by provider: 20 MB on Bedrock, 30 MB on Google-managed Claude, and up to 32 MB on Anthropic-managed cloud services. Oversized first-party requests return 413 `request_too_large`.
- Anthropic
Anthropic Resumes Billing for Claude Refusals Before Output in Three Safeguard Categories
Anthropic resumed billing for Claude requests refused before any output when its safeguards classify them as biology, frontier LLM development, or reasoning extraction. The charge uses the model's normal rates across platforms. Other pre-output refusals remain unbilled, but still count against rate limits.
- Anthropic
Anthropic Open-Sources a Commerce Agents Blueprint for Claude
Anthropic's open-source Claude Commerce Agents blueprint, released September 2, 2026, ships a shopping agent and a merchant agent that run on the Messages API, the Agent SDK, or Managed Agents, four runnable verticals, a Claude Code plugin, and an engineering deep-dive. It is Apache-2.0, unmaintained by design, and no demo places an order or charges a card.
- OpenAI
OpenAI Says Its Developers Plugin Now Has a Showcase Tab
OpenAI Developers says its plugin now has a Showcase tab with starter projects for agents, creative tools, data visualization, and voice. Separately, OpenAI's public API demos page lists working examples and shows that Codex and source actions vary by project. The announcement says users can open a project in Codex and build on it.
Corrections and Updates
- October 9, 2026: This page's rate table covered Opus 5, Sonnet 5, Sonnet 4.x, Haiku 4.5, and Fable 5; it now adds Opus 5.5, Sonnet 5.5, and Haiku 5.5 at the same limits, shows Fable 5.x as one bucket for Fable 5.1 and 5, marks Haiku 3.5 retired, and names the button "Request tier increase."
- August 12, 2026: This page first rested on a developer post on X and repeated its "5x higher rate limits" figure. Anthropic's release notes date the change to June 26, 2026 and describe it as Sonnet and Haiku limits matching Opus at every tier; no Anthropic document states a 5x figure.
Sources
Re-verified in full on October 9, 2026:
- Claude Platform docs, "Rate limits": https://platform.claude.com/docs/en/api/rate-limits
- Claude Platform release notes (June 26, 2026 entry): https://platform.claude.com/docs/en/release-notes/overview
Cited as originally published, and not re-readable by an automated check:
- ClaudeDevs developer post (July 2, 2026): https://x.com/ClaudeDevs/status/2072818299361263778 -- the source of this article's original "no longer based on API spend" and "5x higher rate limits" wording. x.com blocks automated fetches, so that wording is reported as published rather than re-confirmed, and the figures above come from Anthropic's documentation instead.
Frequently Asked Questions
What are the Claude API usage tiers now?
Three standard tiers: Start, Build, and Scale, plus a Custom tier arranged with an account team. New organizations may begin in an Evaluation tier with limits below the published standard limits, which rise automatically as usage history builds.
What are the Claude API rate limits per tier?
For Opus 5.5, Sonnet 5.5, Haiku 5.5, and the other current Opus, Sonnet, and Haiku models: Start is 1,000 RPM, 2M input and 400k output tokens per minute. Build is 5,000 RPM, 5M and 1M. Scale is 10,000 RPM, 10M and 2M. Fable 5.x is lower.
Are Claude usage tiers still based on API spend?
Placement is not. Anthropic's docs say organizations are placed on a tier automatically based on usage history and account standing. Each tier still carries a monthly spend cap: $500 on Start, $1,000 on Build, $200,000 on Scale, and none on Custom.
Do cached tokens count toward Claude rate limits?
For most Claude models, no. Only uncached input tokens and cache-writes count toward the input-tokens-per-minute limit; cache reads do not. Anthropic's example: a 2M ITPM limit with an 80% cache hit rate processes about 10M total input tokens per minute.
Where can developers verify their exact RPM and token limits?
The Rate limits page in the Claude Console shows your organization's tier and current limits, and the Rate Limits API returns them programmatically. Response headers report the limit in force, tokens remaining, and the reset time on every call.