General

SEC-Bench Pro leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on SEC-Bench Pro is 85.4% by GPT-6 Astra. 6 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	GPT-6 Astra	85.4%	measured
2	GPT-6.1 Sol	78.8%	measured
3	Claude Mythos 5	76.1%	estimated ± 4.3 pp, medium confidence
4	Claude Mythos Preview	73.3%	estimated ± 4.3 pp, medium confidence
5	GPT-5.6 Sol	71.2%	measured
6	GLM-5.3	67.7%	estimated ± 4.3 pp, medium confidence
7	GPT-5.6 Terra	67.0%	estimated ± 4.3 pp, medium confidence
8	GPT-6 Sol	66.3%	measured
9	MiMo-V2.6-Pro	66.3%	measured
10	DeepSeek V4.1 Flash	62.8%	measured
11	GPT-5.6 Luna	54.9%	estimated ± 4.3 pp, medium confidence
12	Kimi K3	53.9%	estimated ± 4.3 pp, medium confidence
13	MiMo-V2.6-Flash	47.5%	measured
14	GPT-5.5	45.8%	measured
15	GPT-6 Luna	34.2%	measured

All leaderboards

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General