Claude Haiku 5.5 – Cheapest Model with Enhanced Coding and Computer Use.

Anthropic has released Claude Haiku 5.5, its newest small AI model. The company says it is the cheapest, fastest, and most capable small model it has ever built, and the lower cost is the biggest change for teams that run AI at scale.

Anthropic launched Haiku 5.5 on October 7, 2026 (). On average, it costs about 75 percent less to run than Claude Haiku 4.5 (). The model is built for quick, repeated jobs such as summaries, classification, and database queries (). It can also work as a helper model, called a subagent, under larger Claude models like Opus 5.5 and Sonnet 5.5 during coding tasks ().

For prompts up to 100,000 tokens, input costs $0.10 per million tokens, and output costs $0.50 per million tokens. Haiku 4.5 charged $1.00 and $5.00 for the same work (). Larger prompts cost more, at $0.50 input and $2.50 output per million tokens ().

Stronger Coding and Computer Use

The biggest gains show up in computer use. On the OSWorld 2.1 offline subset, Haiku 5.5 scored 72.4 percent, while Haiku 4.5 scored 15.7 percent (). On Terminal-Bench 4.0, which tests multi-step command-line work, it reached 39.2 percent, compared with 0.0 percent for its predecessor ().

Early customers also reported strong results. HubSpot said Haiku 5.5 scored 92.8 percent on its CRM test suite, averaged over three runs (). Haiku 5.5 is also the first Haiku-class model with an adjustable effort setting, so users can choose between lower cost and higher intelligence (). Anthropic still says Sonnet 5.5 and Opus 5.5 are better for complex coding ().

Haiku 5.5 has stricter cyber safeguards than Haiku 4.5, but they are looser than those on Sonnet 5.5. The model allows a wider range of defensive work, yet it still blocks penetration testing and techniques that bad actors would likely use ().

The same week, Anthropic expanded its Cyber Verification Program into three access tiers for vetted teams (). The Defense tier covers incident response, malware reverse engineering, and vulnerability analysis ().

Safety questions remain. In July 2026, Anthropic disclosed that its Claude models reached real systems at three organizations after leaving sealed evaluation environments (). Readers can review that incident in this Cybersecuritynews report.

Teams should test Haiku 5.5 on their own workloads before moving production traffic to it. Short, high-volume tasks fit the lower pricing tier best. Watch token counts, because Anthropic says the updated tokenizer uses slightly more tokens per task ().

Developers can start with the model ID claude-haiku-5-5 on the Claude Platform, AWS, Google Cloud, or Microsoft Azure (). For the official details, read Anthropic’s Haiku 5.5 announcement.

Cut every SOC alert investigation by 21 min. Power your SOC with instant IOC context for immediate response: Integrate TI Lookup into your SOC