Anthropic · Claude 5
Claude Haiku 5.5
Anthropic's current small model (Oct 2026), priced about 90% below Haiku 4.5 for prompts up to 100K tokens and far ahead of it on agentic work.
- AA Intelligence
- 43#21 of 151
- Input / output per 1M
- $0.10/ $0.50
- Output speed
- 236tok/s
- Context window
- 1M tokens
Composite indexes
- Artificial Analysis Intelligence Index#21 of 15143
Leader: Claude Opus 5.5 at 58
Coding
- Terminal-Bench 4.0#20 of 9639.2%33%
Leader: Claude Sonnet 5.5 at 70.6%
- SciCode#25 of 10055%
Leader: Claude Opus 5.5 at 66.9%
Agents and tool use
- GDPval-AA#13 of 1091618
Leader: Muse Wonder Full at 1857
- OSWorld-Verified#14 of 2172.4%
Leader: Kimi K3 at 84.8%
Reasoning and knowledge
- Humanity’s Last Exam#23 of 15345.9%44%
Leader: Claude Sonnet 5.5 at 64.5%
- CritPt#21 of 14419%
Leader: Muse Wonder Full at 33%
Vision and long context
- AA Long Context Reasoning#15 of 14483%
Leader: Kimi K3 at 88.7%
Human preference
- LMArena WebDev#25 of 581587
Leader: Claude Opus 5.5 at 1818
Other published results
| Result | Score | Reported by | Source |
|---|---|---|---|
| AA Intelligence Index (medium) · AA Intelligence Index v4.3.2, medium (the API default); low 29, high 38, xhigh 41 | 34 index | independent | Artificial Analysis: Claude Haiku 5.5 max vs medium ↗ |
| HLE with tools | 57.4% | vendor | Anthropic: Claude Haiku 5.5 ↗ |
| FrontierCode 1.1 (Main) | 46.4% | vendor | Anthropic: Claude Haiku 5.5 ↗ |
| AA-Briefcase v1.1 · max effort | 1577 Elo | independent | Artificial Analysis: Claude Haiku 5.5 (max) ↗ |
| AutomationBench-AA (partial score) · max effort | 35% | independent | Artificial Analysis: Claude Haiku 5.5 (max) ↗ |
| AA-Omniscience Index · max effort | 11 index (-100 to 100) | independent | Artificial Analysis: Claude Haiku 5.5 (max) ↗ |