benchgap
General

FailureSensorIQ leaderboard

As of 2026-11-18, the highest measured score on FailureSensorIQ is 69.6% by GPT-6 Astra. 114 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1GPT-6 Astra69.6%measured
2GPT-6.2 Astra69.4%estimated ± 2.1 pp, medium confidence
3Gemini 3 Pro69.1%measured
4Gemini 3.8 Flash69.0%measured
5Gemini 4 Carbon68.7%estimated ± 1.7 pp, medium confidence
6Gemini 4 Argon68.7%estimated ± 1.7 pp, medium confidence
7GPT-6.1 Sol68.6%estimated ± 1.7 pp, high confidence
8Gemini 3.1 Pro68.6%estimated ± 1.7 pp, high confidence
9Claude Fable 5.268.6%estimated ± 1.7 pp, high confidence
10GPT-5.568.5%estimated ± 1.7 pp, high confidence
11Claude Opus 568.4%estimated ± 1.7 pp, high confidence
12Gemini 3.7 Flash68.3%estimated ± 1.7 pp, high confidence
13GPT-5.6 Sol68.3%estimated ± 1.7 pp, high confidence
14Kimi K368.3%estimated ± 1.7 pp, high confidence
15Kimi K3.1 Pro68.3%estimated ± 1.7 pp, high confidence
16Muse Wonder68.3%estimated ± 1.7 pp, high confidence
17DeepSeek V4.1 Flash 101768.2%estimated ± 1.7 pp, high confidence
18Claude Fable 568.2%estimated ± 1.7 pp, high confidence
19Gemini 3.5 Flash68.2%estimated ± 1.7 pp, high confidence
20Claude Opus 4.6 (Adaptive)68.2%estimated ± 1.7 pp, high confidence
21GPT-5.5 Pro68.2%estimated ± 2.1 pp, high confidence
22GPT-5.468.1%estimated ± 1.7 pp, high confidence
23Claude Opus 4.868.1%estimated ± 1.7 pp, high confidence
24GPT-6 Sol68.0%estimated ± 1.7 pp, high confidence
25Claude Fable 5.168.0%estimated ± 1.7 pp, high confidence
26Grok 4.20 Multi-agent68.0%estimated ± 1.7 pp, high confidence
27Gemini 3.6 Flash68.0%estimated ± 1.7 pp, high confidence
28Claude Opus 5.567.9%estimated ± 1.7 pp, high confidence
29Qwen3.8 Max67.9%estimated ± 1.7 pp, high confidence
30GPT-5.4 Pro67.9%estimated ± 2.1 pp, high confidence
31Kimi K2.667.8%estimated ± 1.7 pp, high confidence
32GPT-5.2-Codex67.8%estimated ± 1.8 pp, low confidence
33GPT-6.1 Terra67.8%estimated ± 1.7 pp, high confidence
34Grok 4.2067.7%estimated ± 1.7 pp, high confidence
35Muse Spark 1.367.7%estimated ± 1.7 pp, high confidence
36Qwen3.7 Max67.7%estimated ± 1.7 pp, high confidence
37Muse Spark 1.167.7%estimated ± 1.7 pp, high confidence
38GPT-5.3 Codex67.6%estimated ± 1.8 pp, medium confidence
39o367.6%measured
40GPT-5.267.6%estimated ± 1.7 pp, high confidence
41Hy467.6%estimated ± 1.7 pp, high confidence
42Gemini 3 Flash67.5%estimated ± 1.7 pp, high confidence
43Claude Sonnet 5.567.4%measured
44Grok 4.667.3%estimated ± 1.7 pp, high confidence
45Grok 4.567.3%estimated ± 1.7 pp, high confidence
46Muse Spark 1.267.2%estimated ± 1.7 pp, high confidence
47GPT-5.6 Terra67.2%estimated ± 1.7 pp, high confidence
48Muse Spark Open67.2%estimated ± 1.7 pp, high confidence
49GLM-5.167.1%estimated ± 1.7 pp, high confidence
50Grok 4.767.1%estimated ± 1.7 pp, high confidence
51Step 5 Preview67.1%estimated ± 1.7 pp, high confidence
52Gemini 2.5 Pro67.0%measured
53Claude Sonnet 566.9%estimated ± 1.7 pp, high confidence
54Qwen3.7 Plus66.9%estimated ± 1.7 pp, high confidence
55GLM-5.266.9%estimated ± 1.7 pp, high confidence
56GLM-5.366.9%estimated ± 1.7 pp, high confidence
57Qwen 3.6 Max (preview)66.9%estimated ± 1.7 pp, high confidence
58GPT-5.4 nano66.8%estimated ± 2.1 pp, high confidence
59Claude Opus 4.566.7%measured
60Gemma 4 31B66.6%estimated ± 1.7 pp, high confidence
61Nemotron 3 Ultra66.5%estimated ± 5.0 pp, medium confidence
62GPT-5.6 Luna66.5%estimated ± 1.7 pp, high confidence
63Claude Opus 4.766.5%estimated ± 2.1 pp, high confidence
64GPT-6 Luna66.5%estimated ± 1.7 pp, high confidence
65Hy4 preview66.4%estimated ± 1.7 pp, high confidence
66Qwen3.5 397B66.4%estimated ± 5.0 pp, medium confidence
67Claude Haiku 5.566.3%estimated ± 1.7 pp, high confidence
68MiniMax M366.2%estimated ± 1.7 pp, high confidence
69GLM-4.766.2%estimated ± 1.8 pp, medium confidence
70Kimi K2.566.2%estimated ± 1.8 pp, medium confidence
71GPT-566.1%estimated ± 2.1 pp, high confidence
72GPT-5.4 mini66.0%estimated ± 1.7 pp, high confidence
73Gemini 3.5 Flash-Lite65.9%estimated ± 1.7 pp, high confidence
74Qwen3.6 Plus65.9%estimated ± 1.7 pp, high confidence
75DeepSeek V4 Flash 073165.8%estimated ± 2.1 pp, high confidence
76Gemini 2.5 Flash65.5%measured
77Grok 4.365.5%estimated ± 1.7 pp, high confidence
78Qwen3.8-27B65.4%estimated ± 1.7 pp, high confidence
79Seed 2.1 Turbo65.3%estimated ± 2.1 pp, high confidence
80Claude Opus 4.5 Thinking65.3%estimated ± 1.7 pp, high confidence
81Qwen3.5-122B-A10B65.2%estimated ± 1.7 pp, high confidence
82GLM-565.2%estimated ± 1.8 pp, medium confidence
83Qwen3.5-27B65.0%estimated ± 1.7 pp, high confidence
84GLM-5.3-Flash64.9%estimated ± 2.1 pp, high confidence
85Inkling64.9%estimated ± 2.1 pp, high confidence
86Seed 2.1 Pro64.9%estimated ± 2.1 pp, high confidence
87o4-mini64.8%measured
88Claude Sonnet 4.664.7%estimated ± 1.7 pp, high confidence
89Qwen3.7 Flash64.6%estimated ± 1.7 pp, high confidence
90Claude Opus 4.664.6%estimated ± 1.7 pp, high confidence
91Grok 4.1 Fast (Reasoning)64.6%measured
92GPT-5.1-Codex-Max64.6%estimated ± 2.8 pp, low confidence
93Qwen3.6-35B-A3B64.5%estimated ± 1.7 pp, high confidence
94Hy364.5%estimated ± 1.7 pp, high confidence
95o3-mini64.4%estimated ± 2.1 pp, high confidence
96Step 3.7 Flash64.3%estimated ± 1.7 pp, high confidence
97Kimi K2.7 Code64.3%estimated ± 2.1 pp, high confidence
98Qwen3.5 Flash64.3%estimated ± 2.1 pp, high confidence
99Claude Sonnet 4.564.2%measured
100Claude Sonnet 4.5 Thinking64.2%estimated ± 1.7 pp, high confidence
101GPT-4.164.1%estimated ± 2.1 pp, high confidence
102Grok 4 Fast (Reasoning)64.1%measured
103o164.0%estimated ± 2.1 pp, high confidence
104Qwen3.6-27B64.0%estimated ± 5.0 pp, medium confidence
105MiMo-V2.5-Pro64.0%estimated ± 1.7 pp, high confidence
106GPT-4o63.9%estimated ± 2.1 pp, high confidence
107Qwen3.5 Plus63.9%estimated ± 2.1 pp, high confidence
108Claude 3 Opus63.6%estimated ± 2.1 pp, medium confidence
109MiniMax M3.1 Flash63.5%estimated ± 1.7 pp, high confidence
110GPT-4o mini63.5%estimated ± 2.1 pp, medium confidence
111Inkling-Small63.5%estimated ± 5.0 pp, medium confidence
112Mistral Large 463.4%estimated ± 1.7 pp, high confidence
113MiMo-V2-Pro63.3%estimated ± 1.7 pp, high confidence
114GPT-4.1 mini63.3%estimated ± 2.1 pp, medium confidence
115MiniMax M2.763.2%estimated ± 1.7 pp, high confidence
116GPT-4.1 nano63.2%estimated ± 2.1 pp, medium confidence
117GPT-5.162.9%measured
118Qwen3 Max62.8%estimated ± 5.0 pp, medium confidence
119Hy3 Preview62.7%estimated ± 1.7 pp, high confidence
120Trinity-Large-Thinking62.6%estimated ± 1.7 pp, high confidence
121Mistral Medium 3.5 128B62.4%estimated ± 1.7 pp, medium confidence
122Grok 4.1 Fast62.3%estimated ± 1.7 pp, medium confidence
123DeepSeek V3.262.0%estimated ± 1.7 pp, medium confidence
124Gemini 3.1 Flash-Lite62.0%estimated ± 1.7 pp, medium confidence
125Llama 4 Maverick62.0%estimated ± 1.7 pp, medium confidence
126Mistral Large 362.0%estimated ± 1.7 pp, medium confidence
127Claude Haiku 4.561.8%measured
128Qwen3 235B 250760.8%measured
129Grok 460.4%estimated ± 1.8 pp, low confidence
130Claude 3.5 Sonnet58.6%measured
131Claude 3.5 Haiku57.6%measured
132Gemini 2.0 Flash55.8%measured
133GPT-OSS 120B50.5%measured
134Mistral Small 3.147.9%measured
135GPT-OSS 20B43.5%measured
136DeepSeek-R1-052829.1%measured

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General