benchmarks.bio

Legacy V0 results

V0 cross-benchmark results

Historical V0 results. See current V1 results. V0 and V1 use different evaluation sets and aggregation rules; scores are not directly comparable.

Overall score is the equally weighted mean of observed benchmark scores.

Data updated 2026-10-09; artifact generated 2026-10-30T21:23:41.641Z. Missing scores are excluded from a model's mean and are never treated as zero.

benchmarks.bio publishes agentic AI benchmark results across three domains: omics, therapeutics, and biosecurity. Every benchmark is built from a snapshot of real experimental data captured immediately before a target analysis step, paired with a deterministic grader that checks whether an agent recovered the actual biological result rather than whether it produced plausible-looking output. Results are reported per model and per harness rather than per model alone, because the harness an agent runs under changes its score substantially. The overall score in the table below is the equally weighted mean of a configuration's observed scores across the 13 capability benchmarks; a missing benchmark score is excluded from that mean and is never counted as zero. V0 and V1 cover different evaluation sets and should not be read as a single time series.

Capabilities leaderboard across 13 benchmarks
RankModelHarnessProviderOverall score
1Claude Sonnet 5.5Claude CodeAnthropic51.3%
2GPT-6 AstraPiOpenAI48.7%
3GPT-6 AstraCodexOpenAI48.5%
4Claude Opus 5Claude CodeAnthropic48.0%
5Claude Opus 5PiAnthropic48.0%
6Claude Sonnet 5.5PiAnthropic48.0%
7Grok 4.6Grok BuildSpaceXAI45.4%
8Claude Opus 5.5PiAnthropic45.0%
9GPT-5.6 SolPiOpenAI43.4%
10Gemini 3.8 FlashPiGoogle43.4%
11Claude Opus 4.8PiAnthropic42.7%
12GPT-5.6 TerraPiOpenAI42.4%
13Grok 4.6PiSpaceXAI42.3%
14Claude Opus 5.5Claude CodeAnthropic41.9%
15GPT-5.6 SolCodexOpenAI41.8%
16Gemini 3.5 FlashPiGoogle41.7%
17DeepSeek V4.1 FlashPiDeepSeek41.6%
18GPT-5.6 TerraCodexOpenAI41.5%
19Grok 4.5PiSpaceXAI41.3%
20Claude Opus 4.8Claude CodeAnthropic41.1%
21Claude Opus 4.7Claude CodeAnthropic39.8%
22GPT-5.5CodexOpenAI38.7%
23GPT-5.5PiOpenAI37.8%
24Claude Sonnet 5PiAnthropic37.5%
25Claude Sonnet 5Claude CodeAnthropic37.0%
26Gemini 3.7 FlashPiGoogle36.5%
27Kimi K3PiMoonshot35.3%
28Grok 4.7PiSpaceXAI33.7%
29Gemini 4 ArgonPiGoogle33.3%
30GPT-5.6 LunaPiOpenAI31.0%
31GPT-6 SolCodexOpenAI30.4%
32GPT-6 SolPiOpenAI30.3%
33GPT-5.6 LunaCodexOpenAI29.7%
34Mistral Large 4.0PiMistral25.3%
35Muse Wonder 1.0PiMeta24.0%
36GPT-6 LunaCodexOpenAI23.5%
37Claude Sonnet 4.6PiAnthropic22.3%
38GPT-6 LunaPiOpenAI21.4%
39GLM 5.3PiZ.ai17.3%
40Claude Sonnet 4.6Claude CodeAnthropic16.8%
41GPT-6.1 TerraPiOpenAI16.0%
42Muse Spark OpenPiMeta12.0%
43GPT-6.1 TerraCodexOpenAI9.3%
44Mercury 3PiInception8.0%
45Beam 1.0PiReflection1.3%

Included benchmarks