Benchmark Almanac

DrugDiscoveryBench

GPT-6 Astra 68.7% leads since Sep 2026 Not in model cards

Agents do early-stage drug discovery work, from target identification to lead optimization, in 82 tasks with one checkable answer each. A judge model grades answers against expert rubrics.

A task counts as solved only at a full rubric score, averaged over three runs with a 120 minute timeout. The judge is GPT-5.4. About 25 entries on the board.

Made by Scale AI Released Jun 2026 Genomics and life sciences Public tasks plus a private test set Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Expert Specialists with years of training, often over hours of work
Scored
Graded by a model A language model scores each response, usually against a rubric.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
scaleapi/DrugDiscoveryBench 35 stars Paper Dataset Leaderboard
19
Models scored on its own leaderboard

GPT-6 Astra leads with 68.7%

Best score per model from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best.

19 models. Scroll sideways for the rest.

  1. 68.7 GPT-6 Astra
  2. 65.4 Gemini 4 Argon
  3. 62.2 Muse Spark 1.3
  4. 60.6 DeepSeek V4.1 Flash 1017
  5. 56.5 GPT-5.6 Sol
  6. 53.7 Claude Opus 5
  7. 51.6 GPT-5.5
  8. 50.0 Claude Sonnet 5
  9. 50.0 Gemini 3.5 Flash
  10. 46.8 Claude Opus 4.8
  11. 41.9 Gemini 3.1 Pro
  12. 36.2 GLM-5.2
  13. 35.3 Kimi K2.7 Code
  14. 31.7 DeepSeek-V4-Pro
  15. 31.3 Claude Sonnet 4.6
  16. 29.3 GPT-5.2
  17. 29.3 Qwen3.7-Max
  18. 27.7 Claude Opus 4.6
  19. 22.8 MiniMax-M3
  1. 1 GPT-6 Astra max 68.7%
  2. 2 Gemini 4 Argon high 65.4%
  3. 3 Muse Spark 1.3 high 62.2%
  4. 4 DeepSeek V4.1 Flash 1017 xhigh 60.6%
  5. 5 GPT-5.6 Sol max 56.5%
  6. 6 Claude Opus 5 max 53.7%
  7. 7 GPT-5.5 xhigh 51.6%
  8. 8 Claude Sonnet 5 max 50%
  9. 9 Gemini 3.5 Flash high 50%
  10. 10 Claude Opus 4.8 max 46.8%
  11. 11 Gemini 3.1 Pro high 41.9%
  12. 12 GLM-5.2 xhigh 36.2%
  13. 13 Kimi K2.7 Code xhigh 35.3%
  14. 14 DeepSeek-V4-Pro xhigh 31.7%
  15. 15 Claude Sonnet 4.6 max 31.3%
  16. 16 GPT-5.2 xhigh 29.3%
  17. 17 Qwen3.7-Max xhigh 29.3%
  18. 18 Claude Opus 4.6 max 27.7%
  19. 19 MiniMax-M3 xhigh 22.8%
#ModelScoreSettingDateSource
1 GPT-6 Astra 68.7% max, mini-SWE-agent Sep 2026 Official leaderboard
2 Gemini 4 Argon 65.4% high, mini-SWE-agent Sep 2026 Official leaderboard
3 Muse Spark 1.3 62.2% high, mini-SWE-agent Sep 2026 Official leaderboard
4 DeepSeek V4.1 Flash 1017 60.6% xhigh, mini-SWE-agent Oct 2026 Official leaderboard
5 GPT-5.6 Sol 56.5% max, Codex Jul 2026 Official leaderboard
6 Claude Opus 5 53.7% max, Claude Code Jul 2026 Official leaderboard
7 GPT-5.5 51.6% xhigh, mini-SWE-agent Apr 2026 Official leaderboard
8 Claude Sonnet 5 50% max, mini-SWE-agent Jun 2026 Official leaderboard
9 Gemini 3.5 Flash 50% high, Gemini CLI May 2026 Official leaderboard
10 Claude Opus 4.8 46.8% max, mini-SWE-agent May 2026 Official leaderboard
11 Gemini 3.1 Pro 41.9% high, Gemini CLI Feb 2026 Official leaderboard
12 GLM-5.2 36.2% xhigh, mini-SWE-agent Jun 2026 Official leaderboard
13 Kimi K2.7 Code 35.3% xhigh, mini-SWE-agent Jun 2026 Official leaderboard
14 DeepSeek-V4-Pro 31.7% xhigh, mini-SWE-agent Apr 2026 Official leaderboard
15 Claude Sonnet 4.6 31.3% max, mini-SWE-agent Feb 2026 Official leaderboard
16 GPT-5.2 29.3% xhigh, Codex Dec 2025 Official leaderboard
17 Qwen3.7-Max 29.3% xhigh, mini-SWE-agent May 2026 Official leaderboard
18 Claude Opus 4.6 27.7% max, Claude Code Feb 2026 Official leaderboard
19 MiniMax-M3 22.8% xhigh, mini-SWE-agent Jun 2026 Official leaderboard

4 models have held number one since its release in Jun 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI Meta
Own leaderboard 4 leaders
GPT-6 Astra 68.7%

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-6 Astra Muse Spark 1.3 GPT-5.6 Sol GPT-5.5 Gemini 3.1 Pro Claude Sonnet 4.6 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-6 Astra Muse Spark 1.3

No lab model card reports it yet

We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark.

More in Biology and chemistry

All 96 in Biology and chemistry
  • SUPERChem In use Models solve 500 expert-written chemistry reasoning problems, offered in multimodal and text-only forms. Expert solution paths also allow a reasoning path score. Tencent Hy4 preview 66.4%
  • GeneBench Pro In use Agents analyze messy simulated genomics and biology datasets to estimate a decision-relevant quantity across 129 problems. Scored automatically against known answers. GPT-6 Astra 37.1%
  • LifeSciBench In use Models complete 750 expert-written life science research tasks, often with attached figures, tables, or sequence files. Graded against expert rubrics as score and pass rate. GPT-6 Astra 60.3%
  • MolecularIQ In use Models answer questions about molecular graphs, such as counting, indexing and constrained generation, and a symbolic checker verifies each answer. Intern-S2-397B 62.4%
  • TOMG-Bench In use Models generate molecules from natural language requests in three tasks: molecule editing, molecule optimization and customized generation. Each output is checked for validity and for meeting the requested properties. Intern-S2-397B 66.8%
  • BABE Fading Models reason from experimental results in peer-reviewed biology papers to draw conclusions, with causal and cross-scale inference. Seed2.0 Lite (0428) 57.9%