Overview
Claude Haiku 4.5 was Anthropic's fastest and most cost-efficient model at its release, delivering near-frontier intelligence for real-time and high-volume applications. Released October 15, 2025, it matches Claude Sonnet 4's performance on coding, computer use, and agent tasks at a fraction of the cost.
It is the first Haiku model to include extended thinking, computer use, and context awareness, with controllable reasoning depth and a 128K thinking budget. It supports a 200K-token context window, up to 64K output tokens, and accepts both text and image input.
Pricing is $1 per million input tokens and $5 per million output tokens, with cache reads at $0.10 per million. Its speed and price make it well suited to sub-agents, parallelized execution, and scaled deployment.
| Released | 2025-10-15 |
|---|---|
| License | Proprietary |
| Weights | API only |
| Parameters | Undisclosed |
| Context | 200K |
| Max output | 64K |
| Knowledge cutoff | Feb 2025 |
| Modalities | Text, Vision |
| Status | Generally available |
Benchmarks
- SWE-bench Verified73.3%
- Terminal-Bench41.75%
Scores on a 0–100 scale (25-point gridlines); higher is better. Each benchmark links to its published source.
Pricing
| Input | $1.00 / 1M tokens |
|---|---|
| Cached input | $0.10 / 1M tokens |
| Output | $5.00 / 1M tokens |
Strengths
- Fastest Claude model with the lowest per-token price ($1/$5)
- Coding performance matching Claude Sonnet 4 at about one-third the cost
- Strong agentic coding (73.3% on SWE-bench Verified)
- First Haiku model with extended thinking and computer use
- Ideal for high-throughput sub-agents and parallelized execution
Best for
- Reach for it when you need low-latency, high-volume inference such as real-time chat or large fan-out agent fleets.
- Reach for it as a cost-efficient sub-agent or worker model inside a larger Opus/Sonnet-orchestrated system.
- Reach for it for coding and computer-use tasks where speed and cost matter more than absolute frontier capability.
How to access
| Provider | Model ID |
|---|---|
| Anthropic API ↗ | claude-haiku-4-5-20251001 |
Claude Haiku — every version
The full lineage of the Claude Haiku line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| Claude Haiku 5.5current | 2026-10-07 | 1M | Proprietary |
| Claude Haiku 4.5 | 2025-10-15 | 200K | Proprietary |
| Claude 3.5 Haiku | 2024-10-22 | — | Proprietary |
| Claude 3 Haiku | 2024-03-13 | — | Proprietary |
FAQ
How much does Claude Haiku 4.5 cost?
Claude Haiku 4.5 costs $1 per million input tokens and $5 per million output tokens, making it Anthropic's cheapest current model. Prompt-cache reads are $0.10 per million tokens, and batch processing offers a further 50% discount on both input and output.
What is the context window of Claude Haiku 4.5?
Claude Haiku 4.5 supports a 200K-token context window and can generate up to 64K output tokens per request. It is the first Haiku model to add extended thinking, with a thinking budget of up to 128K tokens. Its knowledge cutoff is February 2025.
Is Claude Haiku 4.5 good for coding?
Yes. Claude Haiku 4.5 scores 73.3% on SWE-bench Verified, matching Claude Sonnet 4's coding ability at roughly one-third the cost and more than twice the speed, which makes it strong for high-volume coding agents and sub-agents.