Chess with an engine
loads medium
Tech. quality100
UX90
Time8 min 47 s
Cost≈ $0.086
Compare →
Claude Haiku 5.5
Loads without errors5/5#1 of 26
Tech. quality98/100#3 of 26
UX92/100#2 of 26
Total cost≈ $0.46
Average time13 min 42 s#23 of 26
Total tokens1,005,674
API price: $0.1 / $0.5 per 1M input / output tokens · as of 06/10/2026 source ↗
What is Claude Haiku 5.5 good for?
The fastest and cheapest Claude, built for high-volume, latency-sensitive tasks such as classification, extraction and routing.
Description based on the maker · source ↗
- Maker
- Anthropic
- Released
- 07/10/2026
- Context window
- 1M tokens
- Model weights
- closed, the model is available through the API only
- Reasoning levels
- low medium high max
How did Claude Haiku 5.5 do in the AI Lab tests?
- 5 of 5 outputs load without errors.
- Its best technical result was on the “Chess with an engine” prompt (100/100).
- It implemented the chess rules flawlessly – passing the objective perft test on all 6 positions.
- Average time per prompt 13 min 42 s#23 of 26
- Average cost per prompt ≈ $0.091#10 of 26
- Technical quality 98/100#3 of 26
- Automatic UX rating 92/100#2 of 26
Outputs
All outputs of Claude Haiku 5.5 on the 5 prompts every other model got as well.
GTA-style game
loads medium
Tech. quality100
UX100
Time31 min 13 s
Cost≈ $0.11
Compare →
Solar System
loads medium
Tech. quality94
UX91
Time11 min 27 s
Cost≈ $0.10
Compare →
New York & Central Park map
loads medium
Tech. quality98
UX78
Time9 min 52 s
Cost≈ $0.090
Compare →
Monitor test
loads medium
Tech. quality97
UX100
Time7 min 13 s
Cost≈ $0.070
Compare →
Cost = tokens priced with the API price list valid on the date shown next to the cost (prices change). ≈ = estimate (only the total token count is known), no ≈ = exact from input / output tokens. “Loads” = the page opens without errors; technical quality also checks behaviour during use – judge the quality in the viewer.