Anthropic Launches Claude Haiku 5.5, Cutting The Cost Of Its Smallest Model By Around 75%: All Details

Anthropic has released Claude Haiku 5.5, which it calls its cheapest, fastest and most capable small model to date. The company says the model costs around 75 percent less to run, on average, than Haiku 4.5, and is available now on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure.

The launch is as much about economics as capability. Small models handle the bulk of repetitive AI workloads, so price per task decides what businesses can afford to automate. Anthropic says Haiku 5.5 costs 90 percent less than Haiku 4.5 for prompts up to 100,000 tokens, which made up about 90 percent of requests to the earlier model. Above that threshold, the cut is 50 percent. Input tokens now cost $0.10 per million and output tokens $0.50 per million for prompts up to 100,000 tokens. Anthropic notes that an updated tokenizer means the model uses slightly more tokens per task, a factor already built into its 75 percent average.

Who it is for?

Anthropic positions Haiku 5.5 for high-volume, cost-sensitive work such as summarisation, compaction, database queries and classification. It is also meant to act as a subagent alongside Opus 5.5 and Sonnet 5.5 on coding tasks, and to suit speed-sensitive uses like live customer support and browser use. The company is candid about the limits: for complex agentic coding, Sonnet 5.5 and Opus 5.5 remain the better choices.

Key features and performance

Haiku 5.5 is the first Haiku-class model with an adjustable effort setting, letting users trade cost against intelligence. On Anthropic's own benchmarks, it scored 72.4 percent on the OSWorld 2.1 offline subset (computer use), against 15.7 percent for Haiku 4.5 and 48.9 percent for OpenAI's GPT-6 Luna. On Terminal-Bench 4.0, it scored 39.2 percent, against 0.0 percent for its predecessor and 16.4 percent for GPT-6 Luna. Sonnet 5.5 scored 83.9 percent and 70.6 percent respectively on the same tests. These figures are company-reported.

Early customers cited speed gains. Asana reported over 30 percent lower latency and up to 2.5x faster inference per agent turn, while Box said Haiku 5.5 scored 11 points higher than Haiku 4.5 at about half the latency.

Haiku 5.5 safeguards

Haiku 5.5's cybersecurity safeguards are stricter than Haiku 4.5's but looser than those on other recent models. They permit more defensive work than Sonnet 5.5's but still block penetration testing. Its biology safeguards match those on Sonnet 5, Sonnet 5.5 and Opus 5.

Anthropic is also halving Sonnet 5.5's cache-read price to $0.10 per million tokens, which it says lowers costs on most agentic work by around 20 percent. Max and Team subscribers will get a monthly API credit: $100 for Max 5x, $200 for Max 20x, and up to $500 pooled for Team plans. The Python and TypeScript SDKs also add beta support for computer use and browser use. Developers can access the model as `claude-haiku-5-5`.