Anthropic’s fast, cheap small model for high-volume work such as subagents, summarising and compaction rather than lead coding. It has a 1M tokens context window. Through Anthropic’s API it costs $0.10 (≤100K) · $0.50 (>100K) per million input tokens and $0.50 (≤100K) · $2.50 (>100K) per million output tokens.
Among the tools we track, it’s available in Claude Code, Copilot, Cursor, Kiro, Amp and Devin. Pricing is tiered: prompts over 100K tokens cost five times more. It’s Claude Code’s default Haiku on the Anthropic API.
What Anthropic reports
- Anthropic reports 39.2% on Terminal-Bench 4.0
- Anthropic reports 46.4% on FrontierCode 1.1 (Main)
Vendor-reported figures, not independently verified by Code×Radar.
Related changes
- Copilot will drop six models, including Claude Fable 5, on December 14 · November 13, 2026
- Claude Haiku 5.5 becomes Claude Code's default Haiku model · October 7, 2026
Sources
- platform.claude.com: overview
- platform.claude.com: pricing
- anthropic.com: claude haiku 5 5
- docs.github.com: supported models
Facts last checked November 15, 2026.