Claude Haiku 5.5
API ID claude-haiku-5-5
Launched October 7, 2026
Claude Haiku 5.5 is an Anthropic model, released October 2026. It has a 1,000,000-token context window and up to 128,000 output tokens. It takes image input and supports reasoning. Standard price: from $0.10 input and from $0.50 output per 1M tokens.
- Status
- Active
- Released
- October 2026
- Shutdown
- not before Oct 7, 2027
- Replacement
- —
claude-haiku-5-5 is active. Anthropic says it won't be retired before October 7, 2027.
Benchmarks
Scores as each source published themComparing with Claude Haiku 4.5, the previous model in this line. A score is compared only with the same benchmark from the same source, at the same setting where both have it. Full comparison: prices, limits, features
Arena
| Arena | Claude Haiku 5.5 | Claude Haiku 4.5 | Change | Rank | Date | Source |
|---|---|---|---|---|---|---|
| WebDev | 1585 ±12 | 1330 ±5 | +255 | 31 of 142 2,771 votes | Oct 13, 2026 | source |
| Agent | 0.019 ±0.010 | not rated | 27 of 54 21,839 sessions | Oct 14, 2026 | source |
Rated as claude-haiku-5.5-high (high effort); Claude Haiku 4.5 as claude-haiku-4-5-20251001. Some were rated at different settings; ratings on one arena share one scale, so the difference still reads directly. WebDev ratings come from people voting between two anonymous answers. Agent is Arena's separate score for real agent sessions, on its own scale. Not rated on Text, Text: Coding and Document, where Claude Haiku 4.5 is. Data: Arena (arena.ai), CC BY 4.0.
Epoch AI
| Benchmark | Claude Haiku 5.5 | Claude Haiku 4.5 | Change | Setting | Run | Source |
|---|---|---|---|---|---|---|
| FrontierMath Tier 4 | 46.3% | not run | max effort | Oct 7, 2026 | source | |
| FrontierMath Tiers 1-3 | 75.1% | not run | max effort | Oct 7, 2026 | source | |
| GPQA Diamond | 89.6% | 71.2% | +18.4 pts | max effort (vs 32k) | Oct 9, 2026 | source |
OTIS Mock AIME 2024-2025Other settings
| 98.9% | 66.7% | +32.2 pts | xhigh effort (vs 32k) | Oct 7, 2026 | source |
| SimpleQA Verified | 23.8% | 13.2% | +10.6 pts | max effort (vs no setting) | Oct 9, 2026 | source |
Run independently by Epoch AI. Data: Epoch AI, Capabilities & benchmarking (epoch.ai/benchmarks), CC BY.
On cloud platforms
All platform dates| Platform | Platform ID | State | Shutdown | Time left | Replacement | Source |
|---|---|---|---|---|---|---|
| Amazon Bedrock | anthropic.claude-haiku-5-5 | Active | not before Oct 7, 2027 | — | source |
Read daily from each platform's own pages. The shutdown date and countdown on this page use Anthropic's own date.
Features
Compare models| Feature | Value | Since | Source |
|---|---|---|---|
| Prompt cachingA prompt prefix you send again and again (system prompt, documents, tools) is cached, so later requests cost less and start faster. | Supported | first seen October 8, 2026 | source |
| Batch APISend many requests as one job that runs in the background, usually finished within 24 hours, at a lower price than live requests. | Supported | first seen October 8, 2026 | source |
| Image inputImages such as screenshots, photos and charts can go in the prompt for the model to read. | Supported | first seen October 8, 2026 | source |
| Reasoning / thinkingThe model can work through a problem step by step before it answers: usually better answers to hard questions, at the cost of more tokens and time. | Supported (thinking: adaptive; default effort: medium) | first seen October 8, 2026 | source |
| temperatureSets how random replies are: lower is more predictable, higher is more varied. Accepted: it works. Ignored: the request succeeds but the setting has no effect. Rejected: the request fails with an error. | Rejected (non-default value returns a 400 error) | first seen October 8, 2026 | source |
| top_pLimits each next word to the most likely choices, another way to make replies more or less varied. Accepted: it works. Ignored: the request succeeds but the setting has no effect. Rejected: the request fails with an error. | Rejected (non-default value returns a 400 error) | first seen October 8, 2026 | source |
| Context windowThe most the model can work with in one request, in tokens: your prompt, conversation history and files. A token is about three-quarters of an English word. | 1,000,000 tokens | first seen October 7, 2026 | source |
| Max output tokensThe longest reply the model can write in one response, in tokens. A token is about three-quarters of an English word. | 128,000 tokens | first seen October 7, 2026 | source |
Not published by Anthropic: Structured outputs, Function calling, PDF input, Fine-tuning, Built-in web search, logprobs.
Price history
USD per 1M tokens, standard tierWe started recording this model's prices on Oct 8, 2026. The history grows with each daily check.
Show as a table
| From | Type | Price |
|---|---|---|
| Oct 8, 2026 | Input | $0.10 |
| Oct 8, 2026 | Output | $0.50 |
Prices
USD per 1M tokens, every type and band, by tier| Type | Band | Price $/1M | In effect |
|---|---|---|---|
| Standard | |||
| input | up to 100k prompt | $0.10 | |
| input | over 100k prompt | $0.50 | |
| cached input | up to 100k prompt | $0.01 | |
| cached input | over 100k prompt | $0.05 | |
| cache write (5 min) | up to 100k prompt | $0.125 | |
| cache write (5 min) | over 100k prompt | $0.625 | |
| cache write (1 hour) | up to 100k prompt | $0.20 | |
| cache write (1 hour) | over 100k prompt | $1.00 | |
| output | up to 100k prompt | $0.50 | |
| output | over 100k prompt | $2.50 | |
| Batch | |||
| input | up to 100k prompt | $0.05 | |
| input | over 100k prompt | $0.25 | |
| output | up to 100k prompt | $0.25 | |
| output | over 100k prompt | $1.25 | |
Changelog
Changes we detected on Anthropic's pages- Feature listed
claude-haiku-5-5: top_p parameter rejected (non-default value returns a 400 error)
Info
- Feature listed
claude-haiku-5-5: temperature parameter rejected (non-default value returns a 400 error)
Info
- Feature listed
claude-haiku-5-5: Image input supported
Info
- Feature listed
claude-haiku-5-5: Prompt caching supported
Info
- Feature listed
claude-haiku-5-5: Batch API supported
Info
- Feature listed
claude-haiku-5-5: Reasoning / thinking supported (thinking: adaptive; default effort: medium)
Info
- Platform date set
Amazon Bedrock set a shutdown date for claude-haiku-5-5: not before Oct 7, 2027
Info
- Shutdown date set
Anthropic set a shutdown date for claude-haiku-5-5: not before Oct 7, 2027
Info
- Feature listed
claude-haiku-5-5: Max output tokens 128,000 tokens
Info
- Feature listed
claude-haiku-5-5: Context window 1,000,000 tokens
Info
- Added
Anthropic added claude-haiku-5-5
Info
Related news
- ReleasesInfoAnthropic added claude-haiku-5-5
- ReleasesHeads-upWe've launched Claude Haiku 5.5 (claude-haiku-5-5), our most capable model tuned for high-volume and latency-sensitive work.
- PricingInfoWe've lowered the price of prompt cache reads on Claude Sonnet 5.5 from $0.20 USD to $0.10 USD per million tokens: 0.05x the base input price instead of 0.1x.
- SDKsInfov1.12.0
Email alerts
Get an email if claude-haiku-5-5 changes
We'll email you the same morning we spot a deprecation notice, a new or moved shutdown date, a price change, or the model leaving Anthropic's model list.
No password: one email to confirm, and you can unsubscribe any time. Privacy
Last checked · Sources: Anthropic models, Anthropic deprecations, Anthropic pricing, Anthropic model pages