benchgap
Knowledge & reasoning

C-Eval leaderboard

As of 2026-11-21, the highest measured score on C-Eval is 94.0% by Kimi K2.5. 266 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Grok 4.698.8%estimated ± 1.0 pp, low confidence
2Kimi K398.6%estimated ± 1.0 pp, low confidence
3Muse Wonder98.6%estimated ± 1.0 pp, low confidence
4Qwen4 Max98.4%estimated ± 1.0 pp, low confidence
5Claude Opus 4.898.1%estimated ± 1.0 pp, low confidence
6Claude Opus 4.797.7%estimated ± 1.0 pp, low confidence
7Muse Spark 1.297.4%estimated ± 1.0 pp, low confidence
8DeepSeek V4.1 Flash 101797.4%estimated ± 1.0 pp, low confidence
9Gemini 3.7 Flash97.3%estimated ± 1.0 pp, low confidence
10GPT-5.497.2%estimated ± 1.0 pp, low confidence
11Grok 4.597.2%estimated ± 1.0 pp, low confidence
12GPT-5.597.0%estimated ± 1.0 pp, low confidence
13Claude Sonnet 597.0%estimated ± 1.0 pp, low confidence
14Claude Haiku 5.596.5%estimated ± 1.0 pp, low confidence
15Muse Ember96.3%estimated ± 1.0 pp, low confidence
16Grok 4.796.2%estimated ± 1.0 pp, low confidence
17Gemini 3.6 Flash95.7%estimated ± 1.0 pp, low confidence
18Muse Spark 1.195.7%estimated ± 1.0 pp, low confidence
19GLM-5.295.7%estimated ± 1.0 pp, low confidence
20Qwen3.8 Max Preview95.6%estimated ± 1.0 pp, low confidence
21Hy395.6%estimated ± 1.1 pp, low confidence
22Gemini 3.5 Flash95.2%estimated ± 0.9 pp, low confidence
23Claude Opus 4.6 (Adaptive)95.1%estimated ± 1.0 pp, low confidence
24GLM-5.395.0%estimated ± 1.0 pp, low confidence
25Nemotron 3.5 Bolt 74B-A9B95.0%estimated ± 1.1 pp, low confidence
26Muse Spark94.9%estimated ± 1.0 pp, low confidence
27Gemini 3.1 Pro94.9%estimated ± 1.2 pp, low confidence
28GPT-5.4 nano94.8%estimated ± 0.9 pp, low confidence
29Gemini 3.8 Flash94.8%estimated ± 1.0 pp, low confidence
30Ling 3.1 Flash94.8%estimated ± 0.9 pp, low confidence
31Claude Fable 5.194.7%estimated ± 1.2 pp, low confidence
32Inkling 2 Small94.7%estimated ± 1.0 pp, low confidence
33GPT-5.294.7%estimated ± 1.0 pp, low confidence
34Muse Spark 1.394.7%estimated ± 1.2 pp, low confidence
35Qwen3.8-Omni-Flash94.6%estimated ± 0.9 pp, low confidence
36Claude Opus 594.6%estimated ± 1.2 pp, low confidence
37Qwen3.8-Flash-Next94.5%estimated ± 0.9 pp, low confidence
38Qwen3.7 Max94.4%estimated ± 1.0 pp, low confidence
39Claude Fable 594.4%estimated ± 1.2 pp, low confidence
40Grok 4.394.3%estimated ± 0.9 pp, low confidence
41DeepSeek V4 Flash 073194.3%estimated ± 0.9 pp, low confidence
42Claude Opus 4.5 Thinking94.3%estimated ± 1.0 pp, low confidence
43Kimi K2.694.3%estimated ± 0.9 pp, low confidence
44Qwen3.7 Plus94.3%estimated ± 0.9 pp, low confidence
45MiniMax M3.1 Flash94.3%estimated ± 1.0 pp, low confidence
46MiMo-V2-Pro94.1%estimated ± 1.0 pp, low confidence
47Step 5 Preview94.1%estimated ± 1.0 pp, low confidence
48GPT-5.2-Codex94.1%estimated ± 1.0 pp, low confidence
49Qwen 3.6 Max (preview)94.1%estimated ± 1.0 pp, low confidence
50Kimi K2.594.0%measured
51GPT-5.3 Codex94.0%estimated ± 1.2 pp, low confidence
52GPT-6.2 Astra93.9%estimated ± 1.2 pp, medium confidence
53Gemini 3 Pro93.9%estimated ± 1.0 pp, low confidence
54GPT-6 Luna93.9%estimated ± 1.0 pp, low confidence
55GPT-6.2 Sol93.8%estimated ± 1.2 pp, medium confidence
56GPT-6 Astra93.8%estimated ± 1.2 pp, medium confidence
57Sakana Fugu93.7%estimated ± 1.2 pp, medium confidence
58Sakana Fugu-Ultra93.7%estimated ± 1.2 pp, medium confidence
59MiMo-V2.6-Flash93.6%estimated ± 1.0 pp, medium confidence
60GPT-5.6 Sol93.6%estimated ± 1.2 pp, medium confidence
61GLM-5-Turbo93.5%estimated ± 1.0 pp, medium confidence
62Kimi K3.1 Pro93.5%estimated ± 1.2 pp, medium confidence
63Claude Mythos 593.5%estimated ± 1.2 pp, medium confidence
64GLM-5.193.5%estimated ± 0.9 pp, low confidence
65Claude Opus 4.693.5%estimated ± 1.0 pp, medium confidence
66Qwen3.8-27B93.4%estimated ± 0.9 pp, medium confidence
67Muse Wonder Full93.4%estimated ± 1.5 pp, low confidence
68Kimi K2.7 Code93.4%estimated ± 1.2 pp, low confidence
69GPT-6.1 Sol93.4%estimated ± 1.5 pp, low confidence
70Mistral Large 493.4%estimated ± 1.0 pp, medium confidence
71MiMo-V2.5-Pro93.3%estimated ± 1.0 pp, medium confidence
72GLM-593.3%estimated ± 0.9 pp, medium confidence
73MiMo-V2.6-Pro93.3%estimated ± 1.5 pp, low confidence
74GPT-6 Sol93.3%estimated ± 1.5 pp, low confidence
75Qwen3.6 Plus93.3%measured
76Interfaze Beta93.3%estimated ± 1.0 pp, low confidence
77GPT-5.6 Terra93.3%estimated ± 1.2 pp, medium confidence
78GPT-6.1 Terra93.3%estimated ± 1.5 pp, low confidence
79Ornith-1.5-397B93.3%estimated ± 1.2 pp, medium confidence
80Qwen3.8 Max93.2%estimated ± 1.2 pp, medium confidence
81Seed 2.1 Pro93.2%estimated ± 1.0 pp, medium confidence
82GPT-5.6 Luna93.2%estimated ± 1.2 pp, medium confidence
83Hy4 preview93.2%estimated ± 1.2 pp, medium confidence
84Muse Spark Open93.0%estimated ± 1.0 pp, medium confidence
85Qwen3.5 397B93.0%measured
86Nemotron 3.5 Super VL 120B A12B93.0%estimated ± 0.9 pp, medium confidence
87GPT-5.193.0%estimated ± 1.0 pp, medium confidence
88Claude Sonnet 4.693.0%estimated ± 1.0 pp, medium confidence
89Nemotron 3.5 Super VL 120B A12B NVFP492.9%estimated ± 0.9 pp, medium confidence
90DeepSeek V4 Pro92.8%estimated ± 1.2 pp, high confidence
91GPT-5.4 mini92.8%estimated ± 1.0 pp, medium confidence
92Ling 3.0 Flash VL92.7%estimated ± 1.1 pp, medium confidence
93MiMo-V2-Omni92.7%estimated ± 1.0 pp, medium confidence
94GPT-5.1-Codex92.7%estimated ± 1.0 pp, medium confidence
95GPT-5.1-Codex-Max92.7%estimated ± 1.0 pp, medium confidence
96MiniMax M2.792.7%estimated ± 0.9 pp, medium confidence
97GLM-5V-Turbo92.6%estimated ± 1.0 pp, medium confidence
98A.X K292.6%estimated ± 0.9 pp, medium confidence
99GPT-592.4%estimated ± 1.0 pp, medium confidence
100Qwen3.5-27B92.4%estimated ± 1.0 pp, medium confidence
101Command A+92.3%estimated ± 1.0 pp, medium confidence
102Grok 492.3%estimated ± 1.0 pp, medium confidence
103Nemotron 3 Ultra92.3%estimated ± 0.9 pp, medium confidence
104Claude Opus 4.592.2%measured
105Gemma 4 31B92.2%estimated ± 0.9 pp, medium confidence
106GPT-OSS 20B92.2%estimated ± 1.1 pp, medium confidence
107o3-pro92.1%estimated ± 1.0 pp, medium confidence
108Ternary Bonsai 2 27B92.0%estimated ± 1.2 pp, high confidence
109Ornith-1.5-35B-A3B91.9%estimated ± 0.9 pp, medium confidence
110Seed 2.1 Turbo91.9%estimated ± 1.0 pp, medium confidence
111Hy3 Preview91.9%estimated ± 0.9 pp, medium confidence
112o391.9%estimated ± 1.1 pp, medium confidence
113Claude 4 Opus Thinking91.7%estimated ± 1.0 pp, medium confidence
114Qwen3.5-122B-A10B91.7%estimated ± 1.0 pp, medium confidence
115MAI-Thinking-191.7%estimated ± 1.2 pp, high confidence
116Ling 3.0 Flash FP891.7%estimated ± 1.2 pp, high confidence
117Grok 4.1 Fast (Reasoning)91.7%estimated ± 1.0 pp, medium confidence
118Trinity-Large-Preview91.7%estimated ± 1.1 pp, medium confidence
119Trinity-Large-Thinking91.7%estimated ± 1.1 pp, medium confidence
120GLM-4.791.6%estimated ± 0.9 pp, medium confidence
121Solar Open 291.6%estimated ± 1.3 pp, medium confidence
122Inkling 2 Mini91.6%estimated ± 1.0 pp, medium confidence
123Claude Sonnet 4.591.6%estimated ± 1.2 pp, high confidence
124Step 3.7 Flash91.4%estimated ± 1.0 pp, medium confidence
125Qwen3.6-27B91.4%measured
126Qwen3.5-35B-A3B91.4%estimated ± 1.0 pp, medium confidence
127MiniMax M391.2%estimated ± 1.0 pp, medium confidence
128Claude 4.1 Opus91.1%estimated ± 1.0 pp, medium confidence
129MiniCPM5-2B91.1%estimated ± 1.1 pp, medium confidence
130Sarvam 30B91.1%estimated ± 1.1 pp, medium confidence
131Grok 4 Fast (Reasoning)90.9%estimated ± 1.0 pp, medium confidence
132Gemini 3 Flash90.9%estimated ± 1.0 pp, medium confidence
133Grok 3 Mini90.9%estimated ± 1.2 pp, high confidence
134Beam 1.0 501B-A23B90.9%estimated ± 1.0 pp, medium confidence
135Claude 3.7 Sonnet Thinking90.9%estimated ± 1.0 pp, medium confidence
136Ling 3.0 Flash90.8%estimated ± 0.9 pp, medium confidence
137Celeris-190.8%estimated ± 1.1 pp, low confidence
138Exaone 4.0 1.2B90.8%estimated ± 1.1 pp, low confidence
139Exaone 4.0 32B90.8%estimated ± 1.1 pp, low confidence
140Gemma 4 E2B90.8%estimated ± 1.1 pp, low confidence
141Granite-4.0-350M90.8%estimated ± 1.1 pp, low confidence
142Granite-4.0-H-1B90.8%estimated ± 1.1 pp, low confidence
143Granite-4.0-H-350M90.8%estimated ± 1.1 pp, low confidence
144LFM2.5-2.6B90.8%estimated ± 1.1 pp, low confidence
145LFM2.5-8B-A1B90.8%estimated ± 1.1 pp, low confidence
146LFM2.5-VL-1.6B-Extract90.8%estimated ± 1.1 pp, low confidence
147Ling 3.0 Tiny90.8%estimated ± 1.1 pp, low confidence
148Llama 3.1 405B90.8%estimated ± 1.1 pp, low confidence
149Llama 3.1 70B90.8%estimated ± 1.1 pp, low confidence
150Llama 3.3 70B90.8%estimated ± 1.1 pp, low confidence
151Llama 4 Maverick90.8%estimated ± 1.1 pp, low confidence
152Llama 4 Scout90.8%estimated ± 1.1 pp, low confidence
153Mercury 2.590.8%estimated ± 1.1 pp, low confidence
154Nemotron 3 Nano 30B90.8%estimated ± 1.1 pp, low confidence
155Nemotron 3 Nano Omni 30B A3B90.8%estimated ± 1.1 pp, low confidence
156Nemotron Ultra 253B90.8%estimated ± 1.1 pp, low confidence
157Phi-490.8%estimated ± 1.1 pp, low confidence
158Phi-4-reasoning90.8%estimated ± 1.1 pp, low confidence
159Qwen3.5-4B90.8%estimated ± 1.1 pp, low confidence
160Qwen3-Omni-30B-A3B-Instruct90.8%estimated ± 1.1 pp, low confidence
161Qwen3-Omni-30B-A3B-Thinking90.8%estimated ± 1.1 pp, low confidence
162Sarvam 105B90.8%estimated ± 1.1 pp, low confidence
163Ultravox v0.6 Llama 3.3 70B90.8%estimated ± 1.1 pp, low confidence
164o4-mini90.6%estimated ± 1.0 pp, medium confidence
165Claude 4 Opus90.5%estimated ± 1.0 pp, medium confidence
166Claude 4 Sonnet90.5%estimated ± 1.0 pp, medium confidence
167MiMo-V2-Flash90.4%estimated ± 1.0 pp, medium confidence
168DeepSeek V3.290.4%estimated ± 1.0 pp, medium confidence
169Qwen3 Max90.2%estimated ± 1.0 pp, medium confidence
170Ripple-65B Preview90.2%estimated ± 1.3 pp, medium confidence
171Claude 3.7 Sonnet90.1%estimated ± 1.0 pp, medium confidence
172o190.1%estimated ± 1.0 pp, medium confidence
173GLM-4.690.0%estimated ± 1.0 pp, medium confidence
174Qwen3.6-35B-A3B90.0%measured
175K-Exaone89.9%estimated ± 1.0 pp, medium confidence
176Grok 3 [Beta]89.8%estimated ± 1.2 pp, high confidence
177Solar Pro 489.8%estimated ± 1.0 pp, medium confidence
178GLM-4.7-Flash89.8%estimated ± 1.2 pp, high confidence
179Grok Code Fast 189.8%estimated ± 1.0 pp, medium confidence
180Ling 2.6 Flash89.8%estimated ± 1.0 pp, medium confidence
181Quasar 438B89.7%estimated ± 1.2 pp, medium confidence
182DeepSeek V3.189.7%estimated ± 1.0 pp, medium confidence
183Apodex 1.189.7%estimated ± 1.0 pp, medium confidence
184Apodex 1.1 Mini89.7%estimated ± 1.0 pp, medium confidence
185LFM2.5-24B-A2B89.6%estimated ± 1.2 pp, high confidence
186Qwen3 235B-A22B (Reasoning)89.6%measured
187Ornith-1.5-9B89.6%estimated ± 0.9 pp, low confidence
188DeepSeek-R189.5%estimated ± 1.0 pp, medium confidence
189Inkling-Small89.4%estimated ± 1.0 pp, medium confidence
190Kimi K289.4%estimated ± 1.0 pp, medium confidence
191GPT-4.189.4%estimated ± 1.0 pp, medium confidence
192Gemma 4 12B89.3%estimated ± 1.0 pp, medium confidence
193Inkling89.3%estimated ± 1.0 pp, medium confidence
194o3-mini89.3%estimated ± 1.0 pp, medium confidence
195o1-pro89.3%estimated ± 1.0 pp, medium confidence
196Qwen3 235B 250789.1%estimated ± 1.0 pp, low confidence
197Gemini 3.5 Flash-Lite89.1%estimated ± 1.0 pp, medium confidence
198o1-preview89.0%estimated ± 1.0 pp, medium confidence
199Grok 4.1 Fast89.0%estimated ± 1.0 pp, medium confidence
200GLM-4.5-Air88.9%estimated ± 1.0 pp, medium confidence
201Gemini 2.5 Pro88.8%estimated ± 0.9 pp, low confidence
202ZAYA1-8B88.8%estimated ± 1.2 pp, high confidence
203GPT-4.1 mini88.6%estimated ± 1.0 pp, medium confidence
204Gemini 2.5 Flash88.5%estimated ± 1.0 pp, medium confidence
205K-EXAONE 2.088.5%estimated ± 0.9 pp, low confidence
206o1-mini88.5%estimated ± 1.0 pp, medium confidence
207DeepSeek V3 032488.5%estimated ± 1.0 pp, medium confidence
208Nemotron 3 Super 100B88.5%estimated ± 0.9 pp, low confidence
209Spark-X2.6-4B88.5%estimated ± 1.2 pp, high confidence
210GPT-4.588.4%estimated ± 1.0 pp, medium confidence
211QwQ-32B88.4%measured
212Mistral Large 388.4%estimated ± 1.0 pp, low confidence
213Magistral Medium88.3%estimated ± 1.0 pp, low confidence
214Mistral Medium 388.3%estimated ± 1.0 pp, low confidence
215Claude 3.5 Haiku88.3%estimated ± 1.0 pp, low confidence
216Gemini 2.0 Flash88.3%estimated ± 1.0 pp, low confidence
217Phi-4-reasoning-plus88.2%estimated ± 1.2 pp, high confidence
218Claude 3 Opus88.2%estimated ± 1.0 pp, low confidence
219DeepSeek V388.1%estimated ± 1.0 pp, low confidence
220DeepSeek-R1-052888.1%estimated ± 0.9 pp, low confidence
221GPT-4o88.1%estimated ± 1.0 pp, low confidence
222DeepSeek R1 Distill Qwen 32B88.1%estimated ± 1.0 pp, low confidence
223Magistral Small88.0%estimated ± 1.0 pp, low confidence
224Gemini 1.5 Pro87.9%estimated ± 1.0 pp, low confidence
225GPT-4.1 nano87.9%estimated ± 1.0 pp, low confidence
226Solar Pro 387.9%estimated ± 1.0 pp, low confidence
227Devstral Small87.9%estimated ± 1.0 pp, low confidence
228Mistral Large 287.8%estimated ± 1.0 pp, low confidence
229Phi-4 Multimodal Instruct87.8%estimated ± 1.2 pp, low confidence
230Gemma 4 26B A4B87.8%estimated ± 0.9 pp, low confidence
231Mistral Small 3.187.7%estimated ± 1.0 pp, low confidence
232Gemini 1.5 Flash87.7%estimated ± 1.0 pp, low confidence
233GPT-4 Turbo87.7%estimated ± 1.0 pp, low confidence
234Solar Pro 287.7%estimated ± 1.0 pp, low confidence
235Nova Pro87.7%estimated ± 1.0 pp, low confidence
236Qwen2.5 Coder 32B Instruct87.6%estimated ± 1.0 pp, low confidence
237Gemini 2.5 Flash-Lite87.6%estimated ± 1.0 pp, low confidence
238GPT-4o mini87.6%estimated ± 1.0 pp, low confidence
239DeepSeek-V2.587.6%estimated ± 1.0 pp, low confidence
240Muse Glimmer 30B87.5%estimated ± 1.0 pp, low confidence
241Qwen3 32B (Reasoning)87.3%measured
242Claude 3 Haiku87.2%estimated ± 1.0 pp, low confidence
243Granite 4.2 30B87.2%estimated ± 1.0 pp, low confidence
244Nemotron 3.5 Lightning 30B A3B NVFP487.2%estimated ± 1.0 pp, low confidence
245Gemini 1.0 Pro87.2%estimated ± 1.0 pp, low confidence
246DeepSeek R1 Distill Llama 70B87.2%estimated ± 1.2 pp, high confidence
247Granite 4.2 8B87.1%estimated ± 1.0 pp, low confidence
248Gemma 3 27B87.0%estimated ± 1.0 pp, low confidence
249Gemma 4 E4B87.0%estimated ± 1.0 pp, low confidence
250Granite 4.2 3B87.0%estimated ± 1.0 pp, low confidence
251Mistral Medium 3.5 128B87.0%estimated ± 1.0 pp, low confidence
252Mistral Small 486.9%estimated ± 1.0 pp, low confidence
253LFM3-1.7B86.6%estimated ± 1.2 pp, high confidence
254North Mini Code86.6%estimated ± 1.0 pp, low confidence
255Qwen3 235B-A22B86.1%measured
256GPT-OSS 120B86.0%estimated ± 0.9 pp, low confidence
257DeepSeek-R1-0528-Qwen3-8B85.9%estimated ± 1.2 pp, high confidence
258LongCat-Flash-Lite-Sparse85.8%measured
259Qwen2.5-Max85.6%estimated ± 1.2 pp, high confidence
260Mellum2.1-12B-A2.5B-Thinking85.3%estimated ± 1.4 pp, high confidence
261Claude 3.5 Sonnet85.3%estimated ± 1.2 pp, high confidence
262DeepSeek R1 Distill Qwen 14B85.2%estimated ± 1.2 pp, high confidence
263Mellum2-12B-A2.5B-Thinking84.7%estimated ± 1.2 pp, high confidence
264ZAYA1-74B-Preview84.6%estimated ± 1.2 pp, high confidence
265Grok 284.1%estimated ± 1.2 pp, high confidence
266MiniMax-Text-0183.5%estimated ± 1.2 pp, medium confidence
267Qwen3 32B83.3%measured
268Grok 2 mini82.1%estimated ± 1.2 pp, medium confidence
269Gemini 2.0 Flash-Lite81.9%estimated ± 1.2 pp, medium confidence
270Qwen2.5-VL-32B79.8%estimated ± 1.2 pp, medium confidence
271Soofi S 30B-A3B78.5%estimated ± 1.2 pp, medium confidence
272Mellum2-12B-A2.5B-Instruct77.1%estimated ± 1.2 pp, medium confidence
273LFM2.5-350M69.8%estimated ± 1.2 pp, medium confidence
274Llama 3.1 8B69.6%estimated ± 1.2 pp, medium confidence
275Phi-3.5-mini69.6%estimated ± 1.2 pp, medium confidence
276Granite-4.0-1B69.2%estimated ± 1.2 pp, medium confidence
277LFM2.5-VL-450M65.1%estimated ± 1.2 pp, medium confidence
278LFM2.5-230M64.8%estimated ± 1.2 pp, medium confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General