benchmarks.bio

Therapeutics · Capabilities

TxBench-Antibody-Discovery results

Agentic benchmark tasks for antibody discovery and development decisions.

Published score: Full benchmark (100 evaluations) · short horizon · data updated 2026-10-09

This page reports published TxBench-Antibody-Discovery scores from benchmarks.bio. TxBench-Antibody-Discovery contains 100 evaluations built from real experimental data, each one graded deterministically against the biological result the original analysis reached, so a run counts as a pass only when the agent recovers that result rather than when it produces 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. Scores use a 0-100 percentage scale and are not comparable across benchmarks; see the linked repository and paper for task construction, grading, and confidence-interval methodology.

Published TxBench-Antibody-Discovery scores
RankModelHarnessProviderScore
1GPT-6 AstraPiOpenAI56.9%
2GPT-6 AstraCodexOpenAI55.2%
3Claude Opus 5Claude CodeAnthropic53.0%
4Grok 4.6PiSpaceXAI51.9%
5Claude Opus 5PiAnthropic51.4%
6Grok 4.6Grok BuildSpaceXAI51.0%
7Gemini 3.7 FlashPiGoogle50.0%
8Claude Opus 4.8PiAnthropic45.6%
9Gemini 3.5 FlashPiGoogle42.6%
10Claude Opus 4.8Claude CodeAnthropic42.0%
11Claude Opus 4.7Claude CodeAnthropic39.0%
12DeepSeek V4.1 FlashPiDeepSeek39.0%
13Grok 4.5PiSpaceXAI38.9%
14Claude Sonnet 5Claude CodeAnthropic37.3%
15Claude Sonnet 5PiAnthropic33.9%
16GPT-5.6 SolPiOpenAI33.8%
17GPT-5.5PiOpenAI33.6%
18GPT-5.6 SolCodexOpenAI33.3%
19GPT-5.5CodexOpenAI31.5%
20GPT-5.6 TerraPiOpenAI31.0%
21GPT-5.6 TerraCodexOpenAI29.7%
22GPT-5.6 LunaPiOpenAI19.4%
23GPT-5.6 LunaCodexOpenAI18.2%

Methodology and provenance

Scores use a 0–100 percentage scale. See Paper PDF, Original publication, Repository for task construction, grading, aggregation, and uncertainty details.