Benchmark Almanac

Agentic BAIM-LLM Evaluation (ABLE)

Gemini 3.1 Pro 84.6% leads since Feb 2026 Not in model cards

Agents run a viral-protein redesign workflow. Nine tasks cover structure retrieval, design strategy, tool selection and biological AI model use. Outputs are auto-scored.

Developed by SecureBio in 2025. Not public. Three of nine tasks have expert baselines. Many frontier closed models refuse all tasks.

Made by SecureBio Hazardous knowledge Tasks kept private Last model added Oct 2026 Scores refreshed 2 Nov 2026
Level
Research frontier Researchers, on open problems or tasks that take days
Scored
Checked automatically A program checks the result against an answer key, unit tests or the expected final state.
Answers
Agent task Many steps in a terminal, browser, desktop, app, game or robot.
Leaderboard securebio.org
18
Entries on its own leaderboard, 16 models we track

Gemini 3.1 Pro leads with 84.6%

Best score per entry from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

18 entries. Scroll sideways for the rest.

  1. 84.6 Gemini 3.1 Pro
  2. 83.6 GPT-5.5 Pro
  3. 82.5 Kimi K2.5
  4. 78.4 Kimi K3
  5. 76.7 GLM-5.2
  6. 74.7 Gemini 3 Pro
  7. 62.5 Nemotron 3 Ultra 550B A55B
  8. 59.9 Inkling
  9. 51.7 gpt-oss-120b
  10. 50.5 Beam 1.0
  11. 49.1 Nemotron 3 Super 120B A12B
  12. 48.1 Kimi K2 Thinking
  13. 45.2 Qwen3-235B-A22B-Thinking
  14. 42.9 Kimi K2.6
  15. 24.2 Nemotron 3 Nano 30B-A3B
  16. 23.8 DeepSeek-V3.2
  17. 0.0 Claude Sonnet 4.6
  18. 0.0 Claude Opus 4.8
  1. 1 Gemini 3.1 Pro high 84.6%
  2. 2 GPT-5.5 Pro xhigh 83.6%
  3. 3 Kimi K2.5 82.5%
  4. 4 Kimi K3 max 78.4%
  5. 5 GLM-5.2 max 76.7%
  6. 6 Gemini 3 Pro high 74.7%
  7. 7 Nemotron 3 Ultra 550B A55B 62.5%
  8. 8 Inkling xhigh 59.9%
  9. 9 gpt-oss-120b 51.7%
  10. 10 Beam 1.0 max 50.5%
  11. 11 Nemotron 3 Super 120B A12B 49.1%
  12. 12 Kimi K2 Thinking 48.1%
  13. 13 Qwen3-235B-A22B-Thinking Jul 2025 45.2%
  14. 14 Kimi K2.6 42.9%
  15. 15 Nemotron 3 Nano 30B-A3B 24.2%
  16. 16 DeepSeek-V3.2 23.8%
  17. 17 Claude Sonnet 4.6 max 0%
  18. 18 Claude Opus 4.8 max 0%
#EntryScoreSettingDateSource
1 Gemini 3.1 Pro 84.6% high Feb 2026 Official leaderboard
2 GPT-5.5 Pro 83.6% xhigh Apr 2026 Official leaderboard
3 Kimi K2.5 82.5% Jan 2026 Official leaderboard
4 Kimi K3 78.4% max Jul 2026 Official leaderboard
5 GLM-5.2 76.7% max Jun 2026 Official leaderboard
6 Gemini 3 Pro 74.7% high Nov 2025 Official leaderboard
7 Nemotron 3 Ultra 550B A55B 62.5% Official leaderboard
8 Inkling 59.9% xhigh Jul 2026 Official leaderboard
9 gpt-oss-120b 51.7% Aug 2025 Official leaderboard
10 Beam 1.0 50.5% max Oct 2026 Official leaderboard
11 Nemotron 3 Super 120B A12B 49.1% Official leaderboard
12 Kimi K2 Thinking 48.1% Nov 2025 Official leaderboard
13 Qwen3-235B-A22B-Thinking Jul 2025 45.2% Jul 2025 Official leaderboard
14 Kimi K2.6 42.9% Apr 2026 Official leaderboard
15 Nemotron 3 Nano 30B-A3B 24.2% Dec 2025 Official leaderboard
16 DeepSeek-V3.2 23.8% Dec 2025 Official leaderboard
17 Claude Sonnet 4.6 0% max Feb 2026 Official leaderboard
18 Claude Opus 4.8 0% max May 2026 Official leaderboard

5 models have held number one since 2023

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

Qwen OpenAI Google DeepMind Moonshot AI
Own leaderboard 5 leaders
Gemini 3.1 Pro 84.6%

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%) Gemini 3.1 Pro Gemini 3 Pro gpt-oss-120b Qwen3-235B-A22B-Thinking 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) Gemini 3.1 Pro Gemini 3 Pro

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%