benchgap
Agentic · tools

GDP.pdf leaderboard

As of 2026-11-21, the highest measured score on GDP.pdf is 32.6% by Muse Wonder Full. 159 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Muse Wonder Full32.6%measured
2Kimi K3.1 Pro31.4%measured
3Muse Wonder30.6%measured
4GPT-6.2 Sol30.0%measured
5GPT-6.2 Astra29.8%measured
6GPT-6 Cyber28.7%estimated ± 7.4 pp, low confidence
7GPT-6.1 Terra28.6%measured
8Gemini 4 Carbon28.4%measured
9GPT-6 Astra28.0%estimated ± 2.8 pp, medium confidence
10Gemini 4 Neon28.0%measured
11Gemini 4 Argon27.6%estimated ± 3.1 pp, high confidence
12Gemini 4 Hydrogen27.6%measured
13Claude Mythos 5.227.3%estimated ± 7.4 pp, low confidence
14GPT-6.1 Sol27.0%estimated ± 2.8 pp, medium confidence
15Hy426.7%estimated ± 5.5 pp, low confidence
16Claude Fable 5.126.7%estimated ± 2.8 pp, medium confidence
17GPT-5.6 Cyber26.2%estimated ± 7.4 pp, low confidence
18Claude Opus 5.526.2%measured
19Qwen4 Max26.2%measured
20Claude Sonnet 5.525.8%measured
21Inkling 2 Small25.2%measured
22GPT-6 Sol25.0%estimated ± 2.8 pp, low confidence
23Claude Fable 5.224.2%measured
24Claude Mythos 5.124.0%estimated ± 7.4 pp, low confidence
25DeepSeek V4.1 Flash 101723.3%estimated ± 4.4 pp, high confidence
26GPT-6 Luna22.8%measured
27Muse Ember22.8%measured
28LongCat-2.521.9%estimated ± 4.1 pp, medium confidence
29Claude Mythos Preview21.7%estimated ± 7.4 pp, low confidence
30Muse Spark 1.321.4%estimated ± 4.1 pp, medium confidence
31Claude Opus 521.0%estimated ± 4.4 pp, high confidence
32Claude Haiku 5.520.8%measured
33GPT-5.6 Terra20.7%estimated ± 4.4 pp, high confidence
34GPT-5.3-Codex-Spark20.4%estimated ± 4.7 pp, medium confidence
35Grok 4.2020.4%estimated ± 4.7 pp, medium confidence
36Grok 4.720.0%measured
37GLM-5.3-Flash19.9%estimated ± 3.1 pp, high confidence
38Ultramarine 1 Flash19.8%estimated ± 4.1 pp, medium confidence
39MiMo-V2.6-Pro19.2%measured
40Muse Spark Open19.1%estimated ± 4.1 pp, medium confidence
41Hy4 preview19.0%estimated ± 4.1 pp, medium confidence
42Gemini 3.8 Flash18.9%estimated ± 3.1 pp, high confidence
43MiMo-V2.6-Flash18.7%estimated ± 4.1 pp, medium confidence
44Mistral Large 417.6%measured
45Qwen3.8 Max Preview17.1%estimated ± 4.6 pp, high confidence
46Qwen3.8-Flash-Next17.0%estimated ± 4.1 pp, medium confidence
47Grok 4.617.0%estimated ± 4.6 pp, high confidence
48Ling 3.1 Flash17.0%estimated ± 4.6 pp, high confidence
49Claude Fable 517.0%estimated ± 4.6 pp, high confidence
50GPT-5.6 Sol17.0%estimated ± 4.6 pp, high confidence
51DeepSeek V4 Pro17.0%estimated ± 4.6 pp, high confidence
52Muse Spark 1.217.0%estimated ± 4.6 pp, high confidence
53Claude Sonnet 517.0%estimated ± 4.6 pp, high confidence
54GPT-5.6 Luna17.0%estimated ± 4.6 pp, high confidence
55Claude Opus 4.817.0%estimated ± 4.6 pp, high confidence
56DeepSeek V4 Flash 073117.0%estimated ± 4.6 pp, high confidence
57Gemini 3.7 Flash17.0%estimated ± 4.6 pp, high confidence
58Grok 4.517.0%estimated ± 4.6 pp, high confidence
59GLM-5.217.0%estimated ± 4.6 pp, high confidence
60A.X K217.0%estimated ± 4.6 pp, medium confidence
61Apodex 1.117.0%estimated ± 4.6 pp, high confidence
62Apodex 1.1 Mini17.0%estimated ± 4.6 pp, high confidence
63Celeris-117.0%estimated ± 4.6 pp, medium confidence
64Claude 3.5 Haiku17.0%estimated ± 4.6 pp, medium confidence
65Command A+17.0%estimated ± 4.6 pp, medium confidence
66DeepSeek V317.0%estimated ± 4.6 pp, medium confidence
67DeepSeek V3 032417.0%estimated ± 4.6 pp, medium confidence
68Gemini 2.5 Pro17.0%estimated ± 4.6 pp, medium confidence
69Gemini 3.1 Pro17.0%estimated ± 4.6 pp, medium confidence
70Gemini 3.5 Flash17.0%estimated ± 4.6 pp, high confidence
71Gemini 3.5 Flash-Lite17.0%estimated ± 4.6 pp, medium confidence
72Gemini 3.6 Flash17.0%estimated ± 4.6 pp, high confidence
73Gemma 3 27B17.0%estimated ± 4.6 pp, medium confidence
74Gemma 4 12B17.0%estimated ± 4.6 pp, medium confidence
75Gemma 4 26B A4B17.0%estimated ± 4.6 pp, medium confidence
76Gemma 4 31B17.0%estimated ± 4.6 pp, medium confidence
77Gemma 4 E2B17.0%estimated ± 4.6 pp, medium confidence
78Gemma 4 E4B17.0%estimated ± 4.6 pp, medium confidence
79GLM-4.717.0%estimated ± 4.6 pp, medium confidence
80GLM-5.117.0%estimated ± 4.6 pp, medium confidence
81GPT-4.1 mini17.0%estimated ± 4.6 pp, medium confidence
82GPT-4.1 nano17.0%estimated ± 4.6 pp, medium confidence
83GPT-4o17.0%estimated ± 4.6 pp, medium confidence
84GPT-4o mini17.0%estimated ± 4.6 pp, medium confidence
85GPT-5.117.0%estimated ± 4.6 pp, medium confidence
86GPT-5.4 mini17.0%estimated ± 4.6 pp, medium confidence
87GPT-5.4 nano17.0%estimated ± 4.6 pp, medium confidence
88GPT-OSS 120B17.0%estimated ± 4.6 pp, medium confidence
89GPT-OSS 20B17.0%estimated ± 4.6 pp, medium confidence
90Granite 4.2 30B17.0%estimated ± 4.6 pp, medium confidence
91Granite 4.2 3B17.0%estimated ± 4.6 pp, medium confidence
92Granite 4.2 8B17.0%estimated ± 4.6 pp, medium confidence
93Grok 4.317.0%estimated ± 4.6 pp, medium confidence
94Hy317.0%estimated ± 4.6 pp, medium confidence
95Hy3 Preview17.0%estimated ± 4.6 pp, high confidence
96Inkling17.0%estimated ± 4.6 pp, medium confidence
97Inkling-Small17.0%estimated ± 4.6 pp, medium confidence
98K-Exaone17.0%estimated ± 4.6 pp, medium confidence
99K-EXAONE 2.017.0%estimated ± 4.6 pp, medium confidence
100Kimi K2.617.0%estimated ± 4.6 pp, medium confidence
101Kimi K2.7 Code17.0%estimated ± 4.6 pp, medium confidence
102LFM2.5-2.6B17.0%estimated ± 4.6 pp, medium confidence
103Ling 2.6 Flash17.0%estimated ± 4.6 pp, medium confidence
104Ling 3.0 Flash17.0%estimated ± 4.6 pp, medium confidence
105Ling 3.0 Flash VL17.0%estimated ± 4.6 pp, medium confidence
106Ling 3.0 Tiny17.0%estimated ± 4.6 pp, medium confidence
107Llama 3.3 70B17.0%estimated ± 4.6 pp, medium confidence
108Llama 4 Maverick17.0%estimated ± 4.6 pp, medium confidence
109Llama 4 Scout17.0%estimated ± 4.6 pp, medium confidence
110Mercury 2.517.0%estimated ± 4.6 pp, medium confidence
111MiMo-V2.5-Pro17.0%estimated ± 4.6 pp, medium confidence
112MiMo-V2-Flash17.0%estimated ± 4.6 pp, medium confidence
113MiniCPM5-2B17.0%estimated ± 4.6 pp, medium confidence
114MiniMax M2.717.0%estimated ± 4.6 pp, medium confidence
115MiniMax M317.0%estimated ± 4.6 pp, high confidence
116Mistral Large 317.0%estimated ± 4.6 pp, medium confidence
117Mistral Medium 3.5 128B17.0%estimated ± 4.6 pp, medium confidence
118Mistral Small 3.117.0%estimated ± 4.6 pp, medium confidence
119Mistral Small 417.0%estimated ± 4.6 pp, medium confidence
120Muse Glimmer 30B17.0%estimated ± 4.6 pp, medium confidence
121Muse Spark17.0%estimated ± 4.6 pp, medium confidence
122Nemotron 3.5 Bolt 74B-A9B17.0%estimated ± 4.6 pp, medium confidence
123Nemotron 3.5 Lightning 30B A3B NVFP417.0%estimated ± 4.6 pp, medium confidence
124Nemotron 3.5 Super VL 120B A12B17.0%estimated ± 4.6 pp, medium confidence
125Nemotron 3.5 Super VL 120B A12B NVFP417.0%estimated ± 4.6 pp, medium confidence
126Nemotron 3 Nano 30B17.0%estimated ± 4.6 pp, medium confidence
127Nemotron 3 Nano Omni 30B A3B17.0%estimated ± 4.6 pp, medium confidence
128Nemotron 3 Ultra17.0%estimated ± 4.6 pp, medium confidence
129North Mini Code17.0%estimated ± 4.6 pp, medium confidence
130Quasar 438B17.0%estimated ± 4.6 pp, high confidence
131Qwen3.5-122B-A10B17.0%estimated ± 4.6 pp, medium confidence
132Qwen3.6-27B17.0%estimated ± 4.6 pp, medium confidence
133Qwen3.6-35B-A3B17.0%estimated ± 4.6 pp, medium confidence
134Qwen3.6 Plus17.0%estimated ± 4.6 pp, medium confidence
135Qwen3.7 Max17.0%estimated ± 4.6 pp, medium confidence
136Qwen3.7 Plus17.0%estimated ± 4.6 pp, medium confidence
137Solar Pro 317.0%estimated ± 4.6 pp, medium confidence
138Solar Pro 417.0%estimated ± 4.6 pp, medium confidence
139Step 3.7 Flash17.0%estimated ± 4.6 pp, medium confidence
140Trinity-Large-Preview17.0%estimated ± 4.6 pp, medium confidence
141Trinity-Large-Thinking17.0%estimated ± 4.6 pp, medium confidence
142Ultravox v0.6 Llama 3.3 70B17.0%estimated ± 4.6 pp, medium confidence
143Muse Spark 1.116.9%estimated ± 4.1 pp, medium confidence
144Qwen3.8 Max16.8%estimated ± 4.1 pp, medium confidence
145Kimi K316.7%estimated ± 4.1 pp, medium confidence
146MiniMax M3.1 Flash16.7%estimated ± 3.1 pp, medium confidence
147Atria Dawn Preview16.6%estimated ± 4.1 pp, medium confidence
148Inkling 2 Mini16.5%estimated ± 4.1 pp, medium confidence
149Ling 3.0 Flash Sante16.5%estimated ± 6.1 pp, low confidence
150Claude Opus 4.716.5%estimated ± 4.1 pp, medium confidence
151GPT-5.516.4%estimated ± 4.1 pp, medium confidence
152GPT-5.416.4%estimated ± 4.1 pp, medium confidence
153Claude Sonnet 4.616.4%estimated ± 4.1 pp, medium confidence
154Claude Opus 4.616.4%estimated ± 4.1 pp, medium confidence
155GPT-5.216.4%estimated ± 4.1 pp, medium confidence
156GPT-5.3 Codex16.4%estimated ± 4.1 pp, medium confidence
157Qwen3.8-27B16.4%measured
158Claude Opus 4.516.4%estimated ± 4.1 pp, low confidence
159Claude Sonnet 4.516.4%estimated ± 4.1 pp, low confidence
160GPT-5.1-Codex16.4%estimated ± 4.1 pp, low confidence
161GPT-5.2-Codex16.4%estimated ± 4.1 pp, low confidence
162Claude 4.1 Opus16.4%estimated ± 4.1 pp, low confidence
163Qwen3.5 Plus16.4%estimated ± 4.1 pp, low confidence
164Claude 4 Sonnet16.4%estimated ± 4.1 pp, low confidence
165Claude Haiku 4.516.4%estimated ± 4.1 pp, low confidence
166Gemini 3 Flash16.4%estimated ± 4.1 pp, low confidence
167Gemini 3 Pro16.4%estimated ± 4.1 pp, low confidence
168Kimi K2.516.4%estimated ± 4.1 pp, low confidence
169GPT-516.4%estimated ± 4.1 pp, low confidence
170Ornith-1.5-397B15.6%estimated ± 6.1 pp, low confidence
171Hy4 flash14.7%estimated ± 6.1 pp, low confidence
172Sage-A1.514.6%estimated ± 6.1 pp, low confidence
173dots3-note Preview12.3%estimated ± 6.1 pp, low confidence
174Beam 1.0 501B-A23B11.6%measured
175GLM-5.311.2%measured
176Beam 1.0 501B-A23B FP811.0%estimated ± 6.1 pp, low confidence
177Beam 1.0 501B-A23B NVFP411.0%estimated ± 6.1 pp, low confidence
178GLM-510.3%estimated ± 6.1 pp, low confidence
179Step 5 Preview8.6%estimated ± 3.1 pp, medium confidence
180Agents-A17.7%estimated ± 6.1 pp, low confidence
181Ornith-1.5-35B-A3B0.0%estimated ± 6.1 pp, low confidence
182Spark-X2.6-4B0.0%estimated ± 6.1 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General