SecureBio ABLE Tasks With Baselines Only: leaderboard

SecureBio ABLE Tasks With Baselines Only — Native SecureBio benchmark panel; exact named population. Accuracy includes native refusal/failure conventions and source-reported runs. Raw fractions converted only according to this panel: ABLEFull already uses percent, baseline-only and other capability panels use fractions. Metric: Accuracy (%). Source: https://securebio.org/benchmarks/details/

Metric: Accuracy (%). Source: securebio.org. Saturation forecast: Around April 2027. 20 models tracked. Updated 16 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5.5 ProxHigh95#35
2Kimi K2.573.76#256
3Gemini 3 Pro (Preview)High69.31#113
4Kimi K3Max68.63#56
5GLM-5.2Max67.53#141
6GPT-OSS-120B61.54#703
7Gemini 3.1 Pro (Preview)High57.69#93
8Kimi K2 (Thinking)46.5
9Mistral Large 4High46.24#137
10Inkling SmallxHigh34.49#299
11Nemotron 3 Ultra34.05#428
12InklingxHigh32.16#281
13Beam 1.0Max31.37#206
14Nemotron 3 Super29.83#775
15nemotron-3-nano-30B-a3B28.88#1919

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=securebio-able-tasks-with-baselines-only · How It Works · Data refreshed daily, snapshot 2026-11-16.