benchgap
Coding

PostTrainBench v1 leaderboard

As of 2026-11-18, the highest measured score on PostTrainBench v1 is 34.3% by GLM-5.2. 133 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Opus 5.547.7%estimated ± 1.3 pp, low confidence
2Claude Sonnet 5.547.4%estimated ± 4.6 pp, medium confidence
3Claude Mythos 546.4%estimated ± 4.6 pp, medium confidence
4Claude Fable 5.145.9%estimated ± 1.8 pp, low confidence
5GPT-6 Astra45.7%estimated ± 1.3 pp, low confidence
6MiniMax M3.1 Flash44.8%estimated ± 1.8 pp, low confidence
7Gemini 4 Carbon44.8%estimated ± 1.3 pp, low confidence
8Gemini 4 Argon44.2%estimated ± 1.3 pp, low confidence
9Grok 4.643.7%estimated ± 3.6 pp, medium confidence
10GPT-5.6 Terra43.4%estimated ± 3.6 pp, medium confidence
11Claude Fable 542.9%estimated ± 3.6 pp, medium confidence
12GPT-6.1 Sol42.5%estimated ± 1.3 pp, low confidence
13GPT-5.6 Luna40.5%estimated ± 3.6 pp, medium confidence
14GLM-5.3-Flash40.4%estimated ± 1.3 pp, low confidence
15Kimi K3.1 Pro40.1%estimated ± 1.3 pp, low confidence
16Sakana Fugu-Ultra39.5%estimated ± 4.6 pp, medium confidence
17Claude Opus 539.2%estimated ± 1.3 pp, low confidence
18Hy438.7%estimated ± 1.3 pp, low confidence
19GLM-5.338.0%estimated ± 1.3 pp, low confidence
20Ember-135.4%estimated ± 5.5 pp, low confidence
21Inkling 2 Small35.3%estimated ± 1.3 pp, low confidence
22DeepSeek V4 Flash 073135.3%estimated ± 3.6 pp, medium confidence
23Kimi K334.9%estimated ± 1.3 pp, low confidence
24GPT-5.6 Sol34.4%estimated ± 1.3 pp, low confidence
25Ling 3.1 Flash34.3%estimated ± 4.6 pp, high confidence
26GLM-5.234.3%measured
27Claude Opus 4.834.1%measured
28Muse Spark 1.232.5%estimated ± 3.6 pp, high confidence
29Qwen3.8-27B31.8%estimated ± 3.6 pp, high confidence
30Beam 1.0 501B-A23B31.6%estimated ± 4.6 pp, high confidence
31Beam 1.0 501B-A23B FP831.4%estimated ± 4.6 pp, high confidence
32Qwen3.8 Max31.3%estimated ± 3.6 pp, high confidence
33Hy4 preview31.2%estimated ± 4.6 pp, high confidence
34Beam 1.0 501B-A23B NVFP431.1%estimated ± 4.6 pp, high confidence
35Ornith-1.5-397B30.5%estimated ± 4.6 pp, high confidence
36Mistral Large 429.8%estimated ± 3.1 pp, low confidence
37Qwen3.8-Omni-Flash28.7%estimated ± 4.6 pp, high confidence
38Claude Opus 4.728.6%measured
39Qwen3.8-Flash-Next27.8%estimated ± 4.6 pp, high confidence
40Ornith-1.0-397B27.5%estimated ± 4.6 pp, high confidence
41Muse Wonder27.2%estimated ± 1.3 pp, medium confidence
42Inkling-Small27.1%estimated ± 3.6 pp, high confidence
43Muse Spark 1.126.8%estimated ± 3.6 pp, high confidence
44Ripple-65B Preview26.7%estimated ± 4.6 pp, high confidence
45Sage-A1.526.5%estimated ± 4.6 pp, high confidence
46dots3-note Preview26.3%estimated ± 4.6 pp, high confidence
47Grok 4.525.6%estimated ± 1.3 pp, medium confidence
48Gemini 3.7 Flash25.4%estimated ± 3.6 pp, high confidence
49GPT-5.525.0%measured
50Atria Dawn Preview24.8%estimated ± 4.6 pp, high confidence
51Ornith-1.5-35B-A3B24.8%estimated ± 4.6 pp, high confidence
52Laguna S 2.124.6%estimated ± 4.6 pp, high confidence
53Gemini 3.8 Flash24.4%estimated ± 3.6 pp, high confidence
54Sakana Fugu24.2%estimated ± 4.6 pp, high confidence
55Claude Sonnet 523.9%estimated ± 3.6 pp, high confidence
56Composer 2.523.9%estimated ± 3.6 pp, high confidence
57Gemini 3.6 Flash23.9%estimated ± 3.6 pp, high confidence
58Claude Opus 4.623.2%measured
59Qwen3.7 Plus22.7%estimated ± 4.6 pp, high confidence
60Kimi K2.7 Code22.1%estimated ± 3.6 pp, high confidence
61Gemini 3.1 Pro21.6%measured
62Inkling21.4%estimated ± 3.6 pp, high confidence
63GPT-5.221.4%measured
64Step 3.7 Flash21.4%estimated ± 4.6 pp, high confidence
65GPT-5.420.2%measured
66Claude Opus 4.6 (Adaptive)20.2%estimated ± 3.1 pp, medium confidence
67GLM-520.1%estimated ± 4.6 pp, high confidence
68Claude Sonnet 4.620.0%estimated ± 3.1 pp, medium confidence
69DeepSeek V4 Pro19.8%estimated ± 3.1 pp, medium confidence
70Gemini 3.5 Flash19.7%estimated ± 3.1 pp, medium confidence
71GPT-5.1-Codex-Max19.7%measured
72GPT-5.4 mini19.6%estimated ± 3.1 pp, medium confidence
73Spark-X2.6-4B19.3%estimated ± 4.6 pp, high confidence
74Kimi K2.619.1%estimated ± 3.1 pp, medium confidence
75GLM-5.119.0%estimated ± 3.1 pp, medium confidence
76MiniMax M2.719.0%estimated ± 3.1 pp, medium confidence
77GPT-5.4 nano19.0%estimated ± 3.1 pp, medium confidence
78Qwen3.6 Plus19.0%estimated ± 3.1 pp, medium confidence
79GPT-5.119.0%estimated ± 3.1 pp, medium confidence
80Claude Sonnet 4.5 Thinking19.0%estimated ± 3.1 pp, medium confidence
81Claude Opus 4.5 Thinking19.0%estimated ± 3.1 pp, medium confidence
82Gemini 3 Flash19.0%estimated ± 3.1 pp, medium confidence
83GPT-519.0%estimated ± 3.1 pp, medium confidence
84Muse Spark19.0%estimated ± 3.1 pp, medium confidence
85Qwen3.5 Plus19.0%estimated ± 3.1 pp, medium confidence
86MiniMax M2.519.0%estimated ± 3.1 pp, medium confidence
87GPT-5 mini19.0%estimated ± 3.1 pp, low confidence
88GPT-5.1-Codex19.0%estimated ± 3.1 pp, low confidence
89Claude Haiku 4.5 Thinking19.0%estimated ± 3.1 pp, low confidence
90Grok 4.2019.0%estimated ± 3.1 pp, low confidence
91Qwen3 Max19.0%estimated ± 3.1 pp, low confidence
92GLM-4.619.0%estimated ± 3.1 pp, low confidence
93Grok 4.1 Fast (Reasoning)19.0%estimated ± 3.1 pp, low confidence
94Gemini 2.5 Pro19.0%estimated ± 3.1 pp, low confidence
95Gemini 3.1 Flash-Lite19.0%estimated ± 3.1 pp, low confidence
96Grok 4 Fast (Reasoning)19.0%estimated ± 3.1 pp, low confidence
97Gemini 3.5 Flash-Lite18.2%estimated ± 3.6 pp, high confidence
98MiniMax M318.2%estimated ± 3.6 pp, high confidence
99Gemini 3 Pro18.1%measured
100GPT-5.3 Codex17.8%measured
101MAI-Thinking-117.7%estimated ± 4.6 pp, medium confidence
102GPT-5.2-Codex17.2%measured
103BTL-417.2%estimated ± 5.5 pp, low confidence
104Claude Opus 4.517.1%measured
105MiMo-V2.5-Pro16.9%estimated ± 3.6 pp, high confidence
106MiMo-V2-Pro16.8%estimated ± 5.5 pp, low confidence
107Nemotron 3.5 Super VL 120B A12B16.7%estimated ± 4.6 pp, medium confidence
108Apodex 1.116.5%estimated ± 5.5 pp, low confidence
109Nemotron 3.5 Super VL 120B A12B NVFP416.2%estimated ± 4.6 pp, medium confidence
110Muse Glimmer 30B16.1%estimated ± 4.6 pp, medium confidence
111Claude Sonnet 4.516.0%estimated ± 5.5 pp, low confidence
112Qwen3.5 397B15.7%estimated ± 4.6 pp, medium confidence
113Kimi K2.515.5%estimated ± 4.6 pp, medium confidence
114Qwen 3.6 Max (preview)15.5%estimated ± 3.6 pp, high confidence
115Ornith-1.0-35B15.2%estimated ± 4.6 pp, medium confidence
116Qwen3.6-35B-A3B14.3%estimated ± 4.6 pp, medium confidence
117MiMo-V2-Omni13.7%estimated ± 5.5 pp, low confidence
118Grok 4.313.7%estimated ± 3.6 pp, medium confidence
119Claude 4.1 Opus13.5%estimated ± 5.5 pp, low confidence
120Hy3 Preview13.4%estimated ± 5.5 pp, low confidence
121Step 3.5 Flash13.4%estimated ± 5.5 pp, low confidence
122MiMo-V2.513.2%estimated ± 3.6 pp, medium confidence
123Flint13.2%estimated ± 5.5 pp, low confidence
124MiMo-V2-Flash12.6%estimated ± 5.5 pp, low confidence
125Laguna XS 2.112.3%estimated ± 4.6 pp, medium confidence
126Ornith-1.5-9B12.2%estimated ± 4.6 pp, medium confidence
127Jais-3-MoVA-36B-A4B12.1%estimated ± 5.5 pp, low confidence
128Claude 4 Sonnet12.0%estimated ± 5.5 pp, low confidence
129Qwen3.6-27B12.0%estimated ± 3.6 pp, medium confidence
130MAI-Code-1.1-Flash11.9%estimated ± 5.5 pp, low confidence
131Claude 4 Opus11.8%estimated ± 5.5 pp, low confidence
132Jais-3-32B11.8%estimated ± 5.5 pp, low confidence
133Qwen3.5-27B11.8%estimated ± 5.5 pp, low confidence
134Qwen3.5-122B-A10B11.5%estimated ± 5.5 pp, low confidence
135GLM-4.711.2%estimated ± 3.6 pp, medium confidence
136Laguna XS.211.0%estimated ± 4.6 pp, medium confidence
137Nemotron 3 Ultra10.7%estimated ± 3.6 pp, medium confidence
138Qwen3.7 Max10.5%estimated ± 3.6 pp, medium confidence
139Claude Haiku 4.57.8%estimated ± 3.6 pp, medium confidence
140Mistral Medium 3.5 128B7.5%estimated ± 3.6 pp, medium confidence
141Ornith-1.0-9B7.4%estimated ± 4.6 pp, medium confidence
142Ling 3.0 Flash6.0%estimated ± 3.6 pp, medium confidence
143Qwen3.5 Flash5.1%estimated ± 3.6 pp, medium confidence
144LongCat-Flash-Lite-Sparse5.1%estimated ± 4.6 pp, medium confidence
145Grok 40.0%estimated ± 3.6 pp, medium confidence
146Laguna M.10.0%estimated ± 3.6 pp, medium confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General