Agentic · tools

ExploitGym leaderboard

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

As of 2026-10-07, the highest measured score on ExploitGym is 42.4% by GPT-6 Astra. 107 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	GPT-6 Astra	42.4%	measured
2	Atria Dawn Preview	39.9%	estimated ± 6.1 pp, low confidence
3	GPT-6.1 Sol	35.1%	measured
4	Claude Opus 5	34.4%	estimated ± 6.1 pp, low confidence
5	GPT-5.6 Sol	33.7%	measured
6	GPT-5.5 Pro	32.1%	estimated ± 6.1 pp, low confidence
7	Claude Fable 5	30.5%	estimated ± 10.5 pp, low confidence
8	GPT-5.4 Pro	29.5%	estimated ± 6.1 pp, low confidence
9	Qwen3.8 Max Preview	26.8%	estimated ± 12.2 pp, low confidence
10	Claude Mythos 5	25.3%	estimated ± 6.1 pp, low confidence
11	Muse Spark 1.3	24.9%	estimated ± 5.0 pp, low confidence
12	Claude Fable 5.1	24.0%	estimated ± 5.0 pp, low confidence
13	Claude Opus 5.5	24.0%	estimated ± 5.0 pp, low confidence
14	GPT-5.6 Terra	23.2%	measured
15	Claude Sonnet 5.5	23.2%	estimated ± 5.0 pp, low confidence
16	GPT-6 Sol	22.1%	measured
17	Ornith-1.5-397B	20.8%	estimated ± 6.1 pp, low confidence
18	Muse Spark 1.2	20.3%	estimated ± 12.2 pp, low confidence
19	Grok 4.5	18.2%	estimated ± 12.2 pp, low confidence
20	Gemini 3.7 Flash	18.0%	estimated ± 6.9 pp, low confidence
21	Ling 3.1 Flash	17.8%	estimated ± 9.3 pp, low confidence
22	Fugu Cyber	17.8%	estimated ± 9.3 pp, low confidence
23	MiMo-V2.6-Pro	17.8%	measured
24	Gemini 3.8 Flash Cyber	17.8%	estimated ± 9.3 pp, low confidence
25	Claude Mythos Preview	17.5%	measured
26	Kimi K3	17.5%	estimated ± 5.0 pp, low confidence
27	Gemini 3.5 Flash Cyber	17.5%	estimated ± 9.3 pp, low confidence
28	Gemini 4 Argon	17.3%	estimated ± 5.0 pp, low confidence
29	Gemini 3.8 Flash	16.6%	estimated ± 5.0 pp, low confidence
30	Claude Haiku 5.5	16.4%	estimated ± 5.0 pp, low confidence
31	Grok 4.7	15.8%	estimated ± 5.0 pp, low confidence
32	Grok 4.6	15.5%	estimated ± 10.5 pp, low confidence
33	DeepSeek V4.1 Flash	15.3%	measured
34	Mistral Large 4	15.0%	estimated ± 5.0 pp, low confidence
35	GLM-5.3	15.0%	measured
36	Claude Sonnet 5	14.6%	estimated ± 6.1 pp, low confidence
37	Qwen3.8-27B	14.3%	estimated ± 5.0 pp, low confidence
38	GLM-5.3-Flash	14.0%	estimated ± 5.0 pp, low confidence
39	Step 5 Preview	13.9%	estimated ± 5.0 pp, low confidence
40	Inkling	13.7%	estimated ± 5.0 pp, low confidence
41	Muse Glimmer 30B	13.6%	estimated ± 5.0 pp, low confidence
42	MiniMax M3	13.6%	estimated ± 5.0 pp, low confidence
43	Nemotron 3 Ultra	13.6%	estimated ± 5.0 pp, low confidence
44	GLM-5.2	13.5%	estimated ± 10.3 pp, low confidence
45	GPT-5.5	13.4%	measured
46	Claude Opus 4.8	13.3%	estimated ± 6.1 pp, low confidence
47	GPT-5.6 Luna	12.4%	measured
48	GPT-6 Luna	11.6%	measured
49	Claude Opus 4.6	11.4%	estimated ± 6.1 pp, low confidence
50	DeepSeek V4 Pro 0813	10.4%	estimated ± 6.1 pp, low confidence
51	dots3-note Preview	10.1%	estimated ± 6.1 pp, low confidence
52	Kimi K2.6	9.8%	estimated ± 6.1 pp, low confidence
53	Hy4 preview	9.7%	estimated ± 9.3 pp, low confidence
54	Quasar 438B	7.3%	estimated ± 12.2 pp, low confidence
55	GPT-5.4	6.0%	measured
56	MiMo-V2.6-Flash	6.0%	measured
57	Qwen3.8-Flash-Next	5.2%	estimated ± 6.9 pp, low confidence
58	Qwen3.8 Max	5.2%	estimated ± 6.9 pp, low confidence
59	Gemini 3.6 Flash	4.4%	estimated ± 12.2 pp, low confidence
60	Claude Opus 4.7	3.5%	estimated ± 6.9 pp, low confidence
61	Claude Sonnet 4.6	2.0%	estimated ± 6.9 pp, low confidence
62	Claude Opus 4.5	1.5%	estimated ± 9.3 pp, low confidence
63	GLM-5	1.5%	estimated ± 9.3 pp, low confidence
64	Muse Spark	1.5%	estimated ± 9.3 pp, low confidence
65	Gemini 3.5 Flash	1.1%	estimated ± 12.2 pp, low confidence
66	Muse Spark 1.1	0.8%	measured
67	Qwen3.7 Plus	0.6%	estimated ± 6.9 pp, low confidence
68	Agents-A1	0.0%	estimated ± 6.1 pp, low confidence
69	Agents-A1-4B	0.0%	estimated ± 6.1 pp, low confidence
70	Celeris-1	0.0%	estimated ± 12.2 pp, low confidence
71	Claude Opus 4.7 (Adaptive)	0.0%	estimated ± 6.1 pp, low confidence
72	Command A+	0.0%	estimated ± 12.2 pp, low confidence
73	DeepSeek V3	0.0%	estimated ± 12.2 pp, low confidence
74	DeepSeek V4 Flash 0731	0.0%	estimated ± 6.1 pp, low confidence
75	Gemini 2.5 Pro	0.0%	estimated ± 12.2 pp, low confidence
76	Gemini 3.1 Pro	0.0%	estimated ± 12.2 pp, low confidence
77	Gemini 3.5 Flash-Lite	0.0%	estimated ± 12.2 pp, low confidence
78	Gemma 3 27B	0.0%	estimated ± 12.2 pp, low confidence
79	Gemma 4 31B	0.0%	estimated ± 12.2 pp, low confidence
80	GLM-4.7	0.0%	estimated ± 6.1 pp, low confidence
81	GLM-5.1	0.0%	estimated ± 6.1 pp, low confidence
82	GPT-5.2	0.0%	estimated ± 6.1 pp, low confidence
83	GPT-5.4 mini	0.0%	estimated ± 12.2 pp, low confidence
84	GPT-5.4 nano	0.0%	estimated ± 12.2 pp, low confidence
85	GPT-OSS 120B	0.0%	estimated ± 12.2 pp, low confidence
86	GPT-OSS 20B	0.0%	estimated ± 12.2 pp, low confidence
87	Granite 4.2 8B	0.0%	estimated ± 12.2 pp, low confidence
88	Grok 4.3	0.0%	estimated ± 12.2 pp, low confidence
89	Hy3	0.0%	estimated ± 12.2 pp, low confidence
90	Hy3 Preview	0.0%	estimated ± 12.2 pp, low confidence
91	Inkling-Small	0.0%	estimated ± 6.1 pp, low confidence
92	Kimi K2.5	0.0%	estimated ± 6.1 pp, low confidence
93	Kimi K2.5 (Reasoning)	0.0%	estimated ± 6.1 pp, low confidence
94	Kimi K2.7 Code	0.0%	estimated ± 12.2 pp, low confidence
95	Ling 3.0 Flash	0.0%	estimated ± 6.1 pp, low confidence
96	Ling 3.0 Flash FP8	0.0%	estimated ± 12.2 pp, low confidence
97	Llama 4 Maverick	0.0%	estimated ± 12.2 pp, low confidence
98	Llama 4 Scout	0.0%	estimated ± 12.2 pp, low confidence
99	LongCat-Flash-Lite-Sparse	0.0%	estimated ± 6.1 pp, low confidence
100	MiMo-V2.5-Pro	0.0%	estimated ± 12.2 pp, low confidence
101	MiniMax M2.7	0.0%	estimated ± 12.2 pp, low confidence
102	Mistral Large 3	0.0%	estimated ± 12.2 pp, low confidence
103	Mistral Medium 3.5 128B	0.0%	estimated ± 12.2 pp, low confidence
104	Mistral Small 4	0.0%	estimated ± 12.2 pp, low confidence
105	Mistral Small 4 (Reasoning)	0.0%	estimated ± 12.2 pp, low confidence
106	Nemotron 3.5 Lightning 30B A3B NVFP4	0.0%	estimated ± 6.1 pp, low confidence
107	Nemotron 3 Nano 30B	0.0%	estimated ± 12.2 pp, low confidence
108	Nemotron 3 Super 100B	0.0%	estimated ± 12.2 pp, low confidence
109	Ornith-1.5-35B-A3B	0.0%	estimated ± 6.1 pp, low confidence
110	Ornith-1.5-9B	0.0%	estimated ± 6.1 pp, low confidence
111	Qwen3.5-122B-A10B	0.0%	estimated ± 6.1 pp, low confidence
112	Qwen3.5-27B	0.0%	estimated ± 6.1 pp, low confidence
113	Qwen3.5-35B-A3B	0.0%	estimated ± 6.1 pp, low confidence
114	Qwen3.5 397B	0.0%	estimated ± 6.1 pp, low confidence
115	Qwen3.6-27B	0.0%	estimated ± 12.2 pp, low confidence
116	Qwen3.6-35B-A3B	0.0%	estimated ± 12.2 pp, low confidence
117	Qwen3.7 Max	0.0%	estimated ± 12.2 pp, low confidence
118	Beam	0.0%	estimated ± 6.1 pp, low confidence
119	Solar Pro 4	0.0%	estimated ± 6.1 pp, low confidence
120	Step 3.7 Flash	0.0%	estimated ± 6.1 pp, low confidence
121	Trinity-Large-Preview	0.0%	estimated ± 12.2 pp, low confidence
122	Trinity-Large-Thinking	0.0%	estimated ± 12.2 pp, low confidence

ExploitGym leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General