Claude Haiku 5.5

Claude Haiku 5.5 is Anthropic's fastest and cheapest current model, released on Oct 7, 2026 at $0.10 input and $0.50 output per million tokens for prompts up to 100,000 tokens. It suits classification, extraction, routing and subagents, and is on every Claude plan including Free.

Claude Haiku 5.5 is the smallest and cheapest current Claude model, released on Oct 7, 2026 at $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, rising to $0.50 and $2.50 above that. Anthropic says it costs about 75% less to run on average than Haiku 4.5 and is its fastest model at standard speed (Introducing Claude Haiku 5.5, Anthropic, Oct 7, 2026). The docs recommend it for high-volume, latency-sensitive work such as classification, extraction and routing.

The words Claude Haiku 5.5 in black serif type on an off-white background, set along a thin curved line that runs through three collage blocks: yellow paper on the left, a crossed red and black pattern in the middle and blue paper on the right. — Anthropic's launch image for Claude Haiku 5.5: the model's name in black over a curved line between yellow, red and blue paper blocks. Anthropic

Quick facts

Released	Oct 7, 2026
API id	`claude-haiku-5-5` (Bedrock `anthropic.claude-haiku-5-5`)
Price, prompts up to 100K tokens	$0.10 input, $0.50 output per million tokens
Price, prompts over 100K tokens	$0.50 input, $2.50 output per million tokens
Cache reads	$0.01 (up to 100K), $0.05 (over)
Batch price	$0.05 / $0.25 (up to 100K), $0.25 / $1.25 (over)
Context window	1M tokens (500K in Cowork)
Max output	128K tokens
Thinking	Adaptive; can be turned off on the API at `high` effort or below
Default effort	`medium`
Knowledge cutoff	Jun 2026
In the apps	Every plan, including Free
Retirement	Not before Oct 7, 2027

Sources: Models overview, Pricing, Thinking and Model deprecations, Anthropic docs, read Nov 6, 2026; How large is the context window on paid Claude plans?, Claude Help Center, read Nov 6, 2026.

How is Claude Haiku 5.5 priced?

Haiku 5.5 is the only current Claude model with two price tiers. A request whose prompt is up to 100,000 tokens pays the lower rate; a longer one pays five times as much for every token, and the prompt length counts cache reads and writes too, so a long cached prompt still lands in the higher tier (Pricing, read Nov 6, 2026).

Against Haiku 4.5 ($1 input, $5 output), that is 90% cheaper for prompts up to 100,000 tokens and 50% cheaper above. Anthropic says about 90% of Haiku 4.5 requests fell in the cheaper band, and warns that Haiku 5.5's newer tokenizer uses slightly more tokens for the same task (Introducing Claude Haiku 5.5).

What Claude Haiku 5.5 is good for

Anthropic and its customers name these uses:

Classification, extraction and routing at high volume, as the docs recommend.
Subagents and helpers. Cognition paired Haiku 5.5 as a helper to Claude Opus 5.5 and reports a FrontierCode score of 66.2 for the pair; in Claude Code, /model haiku or a subagent's model: haiku setting uses it.
Customer-facing tools where speed counts. Asana reports over 30% lower latency, and HubSpot 92.8% on its CRM test suite (Introducing Claude Haiku 5.5).

It is also the first Haiku with adjustable effort, from Low to Max. Anthropic itself says Claude Sonnet 5.5 and Opus 5.5 remain better for complex agentic coding, and Haiku 5.5 is slower than Opus models running in fast mode.

Claude Haiku 5.5 benchmarks

Anthropic's launch table:

Benchmark	Haiku 5.5	Haiku 4.5	GPT-6 Luna	Sonnet 5.5
GDPval-AA v2.1 (Elo)	1620	735	1437	1840
AA-Briefcase v1.1 (Elo)	1578	614	1336	1824
OSWorld 2.1 (offline subset)	72.4%	15.7%	48.9%	83.9%
Humanity's Last Exam, with tools	57.4%	18.7%	not given	64.5%
Terminal-Bench 4.0	39.2%	0.0%	16.4%	70.6%
FrontierCode 1.1 Main	46.4%	not given	42.4%	52.1% (xhigh)
Chartography, no tools	46.4%	6.4%	29.1%	61.6%

Source: Introducing Claude Haiku 5.5, Anthropic, Oct 7, 2026.

Independent results are thinner. Haiku 5.5 has no row on the public Terminal-Bench 4.0 leaderboard (Terminal-Bench leaderboard, read Nov 6, 2026). Artificial Analysis scores it 36 at max effort and 30 at high on its Intelligence Index, at $0.16 per task at high and $0.04 at low (read Nov 6, 2026). On Arena it ranks 36th in WebDev at 1589 and 125th in Text Arena at 1425 (tables dated Oct 27, 2026). For the price, those are strong results; for hard tasks, larger models stay well ahead. See Claude Benchmarks.

Where Claude Haiku 5.5 is available

It is on every Claude plan, Free included (Plans and pricing, read Nov 6, 2026), and on the Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry and Claude Platform on AWS. The Claude Help Center says thinking cannot be turned off for Haiku 5.5 in the apps, while the API accepts disabled at high effort or below (Using extended thinking, Claude Help Center, read Nov 6, 2026; Thinking). In Cowork its context window is 500K tokens rather than 1M.

Safety

Its cyber safeguards are stricter than Haiku 4.5's but looser than Sonnet 5.5's, and penetration testing requests are still blocked; its biology safeguards match Sonnet 5, Sonnet 5.5 and Opus 5 (Introducing Claude Haiku 5.5).

Frequently asked questions

How much does Claude Haiku 5.5 cost? Claude Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens when the prompt is up to 100,000 tokens, and $0.50 and $2.50 when it is longer. The Batch API halves both.

Is Claude Haiku 5.5 free? Yes, within limits. Claude Haiku 5.5 is on the Free plan alongside Sonnet 5.5.

Is Claude Haiku 5.5 good for coding? For small, quick coding tasks and as a subagent, yes: Anthropic reports 39.2% on Terminal-Bench 4.0, far above Haiku 4.5. For complex agentic coding Anthropic itself recommends Claude Sonnet 5.5 or Opus 5.5.

What is the Claude Haiku 5.5 context window? Claude Haiku 5.5 accepts 1M tokens on the API and in Claude chat on paid plans, and 500K tokens in Cowork. Prompts over 100,000 tokens cost more on the API.

Is Claude Haiku 5.5 faster than Opus 5.5? At standard speed, yes: Anthropic labels Claude Haiku 5.5 its fastest model. Claude Opus 5.5 in fast mode is faster, at much higher cost.

Sources

Introducing Claude Haiku 5.5, Anthropic, Oct 7, 2026
Models overview, Anthropic docs, read Nov 6, 2026
Pricing, Anthropic docs, read Nov 6, 2026
Thinking, Anthropic docs, read Nov 6, 2026
Model deprecations, Anthropic docs, read Nov 6, 2026
How large is the context window on paid Claude plans?, Claude Help Center, read Nov 6, 2026
Using extended thinking, Claude Help Center, read Nov 6, 2026
Plans and pricing, Claude, read Nov 6, 2026
Terminal-Bench leaderboard, tbench.ai, read Nov 6, 2026
Models leaderboard, Artificial Analysis, read Nov 6, 2026
Text Arena and WebDev Arena, Arena, Oct 27, 2026