AA-Briefcase
1.1 / combined Elo ↗- Claude Haiku 5.5:
- GPT-6 Astra:
Compare models
Find the right model, side by side.
Updated: Data as of:
First model
Second model
Model specifications
| Specifications |  Claude Haiku 5.5Anthropic |  GPT-6 AstraOpenAI |
|---|---|---|
| Provider | ||
| Context sizetokens | Not reported | 1,000,000 |
| Cutoff date | 2026-06 | 2026-04-30 |
| Input costUSD / 1M tokens | $0.50 | $10.00 |
| Output costUSD / 1M tokens | $2.50 | $50.00 |
| Max outputtokens | Not reported | 128,000 |
| LatencyTTFT · seconds | Not reported | 365.127 |
| Speedtokens / sec | Not reported | 70.564 |
Benchmark performance
13 shared benchmark comparisons. Versions and setups matched.
Claude Haiku 5.5GPT-6 Astra
AutomationBench-AA
1.0.6 / AA private 657 ↗Tools enabled- Claude Haiku 5.5:
- GPT-6 Astra:
- Claude Haiku 5.5:
- GPT-6 Astra:
HLE-Diamond
1,000 questions / no tools ↗No tools- Claude Haiku 5.5:
- GPT-6 Astra:
Humanity’s Last Exam
Stable HLE — full set ↗Tools enabled- Claude Haiku 5.5:
- GPT-6 Astra:
LiveBench: Agentic
2.0 / 2026-10-25 ↗- Claude Haiku 5.5:
- GPT-6 Astra:
LiveBench: Browser Use
2.0 / 2026-10-25 ↗- Claude Haiku 5.5:
- GPT-6 Astra:
LiveBench: Coding
2.0 / 2026-10-25 ↗- Claude Haiku 5.5:
- GPT-6 Astra:
LiveBench: Data Analysis
2.0 / 2026-10-25 ↗- Claude Haiku 5.5:
- GPT-6 Astra:
LiveBench: Design
2.0 / 2026-10-25 ↗- Claude Haiku 5.5:
- GPT-6 Astra:
LiveBench: Reasoning
2.0 / 2026-10-25 ↗- Claude Haiku 5.5:
- GPT-6 Astra:
LiveBench: Research
2.0 / 2026-10-25 ↗- Claude Haiku 5.5:
- GPT-6 Astra:
Terminal-Bench
4.0 / mini-swe-agent ↗mini-swe-agent- Claude Haiku 5.5:
- GPT-6 Astra:
