Claude Haiku 5.5
Anthropic · released
Small, fast model for high-volume and latency-sensitive jobs such as classification, extraction, routing, summarising and sub-agent work. It is the first Haiku with adjustable reasoning effort and a 1M-token context.
Best for: Classification and extraction · Sub-agents · Real-time support
- Intelligence Index
- 43.4
- #14 of 38 · at Max effort
- Price per 1M tokens
- $0.10 in $0.50 out
- Blended $0.20 at 3:1
- Context window
- 1M
- Up to 128K output
- Output speed
- 242 tokens/s
- $0.21 measured cost per task
Capability profile
One 0–100 score per area, placing this model between the weakest and strongest tracked model across that area's benchmarks.
- Overall68 · #14 of 38
Based on 2 of 3 benchmarks
- Coding65 · #11 of 39
Based on 4 of 5 benchmarks
- Agents & knowledge work67 · #16 of 40
Based on 5 of 6 benchmarks
- Reasoning & science63 · #14 of 39
Based on 2 of 2 benchmarks
- Knowledge & reliability72 · #14 of 39
Based on 2 of 2 benchmarks
- VisionNot enough data
Based on 0 of 2 benchmarks
Reasoning effort trade-off
The same model at each effort setting. Lower settings give up a few points for a fraction of the cost and wait.
| Setting | Intelligence Index | Cost / task | First answer |
|---|---|---|---|
| Max | 43.4 | $0.21 | 7.2 min |
| xHighSweet spot | 41.2 | $0.12 | 87s |
| High | 37.8 | $0.08 | 28s |
| Medium | 34.5 | $0.05 | 13s |
| Low | 29.4 | $0.02 | 9.9s |
Measured by Artificial Analysis. First answer is the median wait until the first answer token, including thinking time. “Sweet spot” is the cheapest setting that keeps at least 90% of the top score for no more than 60% of its cost.
Benchmark results
All third-party results on record. Rank is among the tracked models that have been tested on that benchmark.
Overall
#14 of 38· Max effort· Artificial Analysis
- Vals Index54.3%
#11 of 26· Max effort· $2.99 per test· Vals AI
- Arena · TextNot tested
Coding
- Terminal-Bench 4.032.8%
#13 of 38· Max effort· Artificial Analysis
- SciCode55.0%
#14 of 38· Max effort· Artificial Analysis
- Arena · WebDevNot tested
- Vibe Code Bench90.4%
#3 of 26· Max effort· Vals AI
- Code Migration45.9%
#11 of 26· Max effort· Vals AI
Agents & knowledge work
- GDPval-AA56.0
#12 of 39· Max effort· Artificial Analysis
- Arena · AgentNot tested
- Finance Agent v254.1%
#13 of 26· Max effort· Vals AI
#16 of 26· Max effort· Vals AI
- Legal Research Bench43.3%
#12 of 26· Max effort· Vals AI
- Tax Agent Bench62.2%
#18 of 26· Max effort· Vals AI
Reasoning & science
- Humanity's Last Exam44.4%
#13 of 39· Max effort· Artificial Analysis
- CritPt18.9%
#14 of 39· Max effort· Artificial Analysis
- ARC-AGI-2saturatingNot tested
- ARC-AGI-3Not tested
- GPQA DiamondsaturatingNot tested
Knowledge & reliability
#15 of 39· Max effort· Artificial Analysis
- AA-LCR82.7%
#13 of 39· Max effort· Artificial Analysis
Vision
- MMMU-ProNot tested
- Arena · VisionNot tested
Reported by Anthropic
Numbers the lab published at launch, run on its own harness and settings. Useful context, but not independently verified and never used in the rankings on this site.
- Terminal-Bench 4.0 · Max effort39.2%
- SWE-bench Pro · Max effort64.8%
- OSWorld 2.1 (offline subset) · partial score72.4%
- Humanity's Last Exam · no tools45.9%
Specifications
- Lab
- Anthropic
- Released
- Oct 7, 2026
- Availability
- Generally available
- Context window
- 1,000,000 tokens
- Max output
- 128,000 tokens
- Accepts
- Text, Images, PDFs
- Knowledge cutoff
- Jun 2026
- Weights
- Closed
- Reasoning control
- On by default · effort low to max
- API model ID
claude-haiku-5-5
Pricing
- Input / 1M tokens
- $0.10
- Cached input / 1M
- $0.01
- Output / 1M tokens
- $0.50
- Blended (3:1)
- $0.20
- Measured cost / task
- $0.21
These rates apply to prompts up to 100K tokens. Longer prompts cost $0.50 in / $2.50 out.
Official sources
Closest alternatives
The models nearest to this one on overall score.
23 benchmarks tracked. See methodology and sources.