Best LLMs for Science — Anthropic
The best Anthropic model for science is Claude Opus 5, scoring 93.2% on GPQA Diamond — graduate-level physics, chemistry and biology. Claude Fable 5 (92.6%) and Claude Opus 4.8 (92.0%) round out the top three.
- 1
claude-opus-5ReasoningToolsJSON+160.7 intel · $5.00/M · 60 t/s93.2%GPQA - 2claude-fable-5ReasoningToolsJSON+159.9 intel · $10.00/M · 82 t/s92.6%GPQA
- 3claude-opus-4.8ReasoningToolsJSON+155.7 intel · $5.00/M · 78 t/s92.0%GPQA
- 4claude-opus-4.7ReasoningToolsJSON+153.5 intel · $5.00/M · 55 t/s91.4%GPQA
- 5claude-sonnet-5ReasoningToolsJSON+153.4 intel · $2.00/M · 95 t/s91.1%GPQA
- 6claude-opus-4.6ReasoningToolsJSON+137.8 intel · $5.00/M · 40 t/s84.0%GPQA
- 7claude-opus-4.5ReasoningToolsJSON+134.7 intel · $5.00/M · 200K ctx81.0%GPQA
- 8claude-sonnet-4.6ReasoningToolsJSON+135.9 intel · $3.00/M · 52 t/s79.9%GPQA
- 9claude-sonnet-4.5ReasoningToolsJSON+129.3 intel · $3.00/M · 1M ctx72.7%GPQA
- 10claude-opus-4ReasoningToolsVision25.5 intel · $15.00/M · 23 t/s70.1%GPQA
- 11claude-sonnet-4ReasoningToolsVision25.5 intel · $3.00/M · 46 t/s68.3%GPQA
- 12claude-haiku-4.5ReasoningToolsJSON+123.7 intel · $1.00/M · 123 t/s64.6%GPQA
- 13claude-3-haikuToolsVision3.9 intel · $0.250/M · 89 t/s37.4%GPQA
How this is ranked
AI models ranked by GPQA Diamond — graduate-level physics, chemistry and biology questions written to be un-Googleable, so the score reflects reasoning from knowledge rather than retrieval. The best large language models for scientific research and hard-science work.
Frequently asked
What is the best Anthropic model for science?
The best Anthropic model for science is Claude Opus 5, scoring 93.2% on GPQA Diamond — graduate-level physics, chemistry and biology. Claude Fable 5 (92.6%) and Claude Opus 4.8 (92.0%) round out the top three.
What's a good alternative to Claude Opus 5?
Claude Fable 5 (92.6%) is the closest alternative on this metric, followed by Claude Opus 4.8 (92.0%). See the full ranking above for the tradeoffs.
How many Anthropic models are there?
modelgrep tracks 17 Anthropic models with live benchmarks, speed, latency and per-provider pricing, led on intelligence by Claude Opus 5. 13 of them qualify for this ranking.