Anthropic has launched Claude Haiku 5.5, its latest compact model aimed at high-volume, cost-sensitive workloads such as summarization, classification, and sub-agent execution. Anthropic claims the new model drops per-token prices by up to 90 percent for prompts under 100,000 tokens compared to Haiku 4.5, though an updated tokenizer consumes slightly more tokens per task.

Haiku 5.5 delivers performance gains over its predecessor across multiple benchmarks, scoring 72.4 percent on OSWorld-2.1 for computer use and 39.2 percent on Terminal-Bench 4.0 for agentic coding. It is also the first Haiku-tier model to feature adjustable reasoning levels, allowing users to tune performance against token cost.

Alongside the model launch, Anthropic cut Sonnet 5.5 cache read prices by 50 percent to $0.10 per million tokens and introduced monthly API credits for subscribers. The aggressive pricing adjustments reflect intensifying competition with rival provider OpenAI’s lower-cost model tiers.

Why it matters

  • Significantly reduces API expenses for high-volume tasks, sub-agents, and background computer-use automation.

  • Provides flexible cost-performance trade-offs by introducing adjustable reasoning levels to Anthropic’s fastest model tier.

Source: the-decoder.com