benchgap
Coding

AA Coding Index leaderboard

As of 2026-11-21, the highest measured score on AA Coding Index is 81.6% by Claude Fable 5.1. 184 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Composer 289.7%estimated ± 11.1 pp, low confidence
2Composer 2 Fast87.9%estimated ± 11.1 pp, low confidence
3Gemini 4 Carbon86.6%estimated ± 1.2 pp, medium confidence
4Claude Fable 5.286.4%estimated ± 1.2 pp, medium confidence
5Claude Opus 5.585.1%estimated ± 1.2 pp, medium confidence
6Claude Sonnet 5.583.8%estimated ± 1.2 pp, medium confidence
7Gemini 4 Hydrogen83.1%estimated ± 1.2 pp, medium confidence
8Gemini 4 Argon82.3%estimated ± 1.2 pp, medium confidence
9Claude Fable 5.181.6%measured
10Muse Wonder80.9%estimated ± 1.2 pp, high confidence
11Claude Haiku 5.579.7%estimated ± 1.2 pp, high confidence
12DeepSeek V4.1 Flash 101779.3%estimated ± 1.2 pp, high confidence
13Gemini 4 Neon79.1%estimated ± 2.3 pp, low confidence
14GPT-6.2 Astra79.0%estimated ± 2.3 pp, low confidence
15GPT-6.2 Sol79.0%estimated ± 2.3 pp, low confidence
16Muse Wonder Full78.9%estimated ± 2.5 pp, medium confidence
17GPT-6.1 Sol78.7%estimated ± 2.3 pp, low confidence
18Grok 4.778.5%estimated ± 1.2 pp, high confidence
19GPT-6 Sol78.3%estimated ± 2.3 pp, low confidence
20Claude Mythos 578.2%estimated ± 5.3 pp, medium confidence
21GPT-6.1 Terra78.0%estimated ± 3.2 pp, low confidence
22Claude Opus 578.0%measured
23Qwen4 Max77.9%estimated ± 2.5 pp, medium confidence
24Hy477.9%estimated ± 2.3 pp, low confidence
25Ember-177.8%estimated ± 2.5 pp, high confidence
26Pareto 26.977.4%estimated ± 2.5 pp, high confidence
27GPT-5.6 Sol77.4%measured
28MiniMax M3.1 Flash77.3%estimated ± 2.5 pp, high confidence
29MAI Code 277.1%estimated ± 2.5 pp, high confidence
30Particle-177.0%estimated ± 2.5 pp, high confidence
31GPT-6 Astra76.9%measured
32SWE-276.9%estimated ± 2.3 pp, low confidence
33SWE-2 Flash76.9%estimated ± 2.3 pp, low confidence
34Grok 4.676.8%measured
35MiMo-V2.6-Pro76.7%estimated ± 2.5 pp, high confidence
36GPT-5.6 Terra76.7%measured
37GPT-6 Luna76.6%estimated ± 2.3 pp, low confidence
38Inkling 2 Small76.6%estimated ± 2.3 pp, low confidence
39ERNIE 5.5 Flash Preview76.5%estimated ± 2.5 pp, high confidence
40Claude Fable 576.5%measured
41Gemini 3.8 Flash76.3%measured
42Kimi K376.2%measured
43Pareto 26.10 Preview76.1%estimated ± 2.5 pp, high confidence
44Gemini 3.7 Flash76.1%measured
45Kimi K3.1 Pro76.0%estimated ± 2.3 pp, low confidence
46Mistral Large 475.9%estimated ± 2.3 pp, low confidence
47Muse Spark 1.375.8%measured
48MiMo-V2.6-Flash75.5%estimated ± 2.5 pp, high confidence
49Step 5 Preview75.5%estimated ± 2.5 pp, high confidence
50GPT-5.574.9%measured
51GLM-5.374.8%measured
52Hy4 flash74.5%estimated ± 2.5 pp, high confidence
53LongCat-2.574.5%estimated ± 2.5 pp, high confidence
54SWE-1.774.5%estimated ± 2.3 pp, low confidence
55Hy4 preview74.4%estimated ± 2.5 pp, high confidence
56Inkling 2 Mini74.3%estimated ± 2.3 pp, low confidence
57Claude Opus 4.874.3%measured
58Claude Opus 4.773.6%measured
59Ultramarine 1 Flash73.5%estimated ± 2.5 pp, high confidence
60GLM-5.3-Flash73.1%estimated ± 1.2 pp, high confidence
61Qwen3.8-Flash-Next73.1%measured
62MAI Code 2 Flash73.0%estimated ± 2.5 pp, high confidence
63Ling 3.1 Flash73.0%estimated ± 2.5 pp, high confidence
64Grok 4.572.5%measured
65Qwen3.8-Omni-Flash72.4%estimated ± 2.5 pp, high confidence
66Muse Spark 1.272.2%measured
67Qwen3.8 Max72.0%estimated ± 2.5 pp, high confidence
68Ornith-1.5-397B71.8%estimated ± 2.5 pp, high confidence
69Qwen3.8 Max Preview71.8%measured
70Claude Sonnet 571.6%measured
71GPT-5.6 Luna71.5%measured
72Muse Spark 1.171.3%measured
73GPT-5.471.1%measured
74Gemini 3.5 Flash70.1%measured
75GPT-5.3 Codex69.4%estimated ± 5.3 pp, medium confidence
76Gemini 3.6 Flash69.2%measured
77DeepSeek V4 Flash 073169.1%measured
78Mercury 368.9%estimated ± 2.5 pp, high confidence
79DeepSeek V4 Pro68.8%measured
80Gemini 3.1 Pro68.8%measured
81GLM-5.268.8%measured
82Ripple-65B Preview68.2%estimated ± 2.5 pp, high confidence
83dots3-note Preview68.1%estimated ± 4.3 pp, high confidence
84Qwen3.8-27B68.1%measured
85Beam 1.0 501B-A23B67.9%estimated ± 2.5 pp, high confidence
86Composer 2.567.9%estimated ± 1.2 pp, medium confidence
87Beam 1.0 501B-A23B FP867.8%estimated ± 2.5 pp, high confidence
88Sakana Fugu-Ultra67.8%estimated ± 5.7 pp, low confidence
89Beam 1.0 501B-A23B NVFP467.7%estimated ± 2.5 pp, high confidence
90Sakana Fugu67.3%estimated ± 5.7 pp, low confidence
91Laguna S 2.167.0%estimated ± 2.5 pp, medium confidence
92Sage-A1.567.0%estimated ± 7.5 pp, medium confidence
93Ornith-1.0-397B66.9%estimated ± 4.3 pp, high confidence
94Muse Spark Open66.3%estimated ± 2.3 pp, low confidence
95Claude Sonnet 4.666.0%estimated ± 3.6 pp, medium confidence
96Qwen3.7 Max66.0%measured
97Seed 2.1 Pro65.8%estimated ± 4.3 pp, high confidence
98Atria Dawn Preview64.8%estimated ± 7.5 pp, medium confidence
99Claude Opus 4.6 (Adaptive)63.2%estimated ± 5.6 pp, medium confidence
100o363.1%estimated ± 11.5 pp, low confidence
101Seed 2.1 Turbo62.3%estimated ± 4.3 pp, high confidence
102Kimi K2.661.8%measured
103Claude Opus 4.561.7%estimated ± 4.3 pp, high confidence
104Ornith-1.5-35B-A3B61.4%estimated ± 2.5 pp, medium confidence
105Qwen 3.6 Max (preview)61.3%estimated ± 4.3 pp, high confidence
106o4-mini61.3%estimated ± 11.5 pp, low confidence
107Quasar 438B61.2%measured
108Claude Opus 4.5 Thinking61.0%estimated ± 5.6 pp, medium confidence
109Gemini 3 Pro61.0%estimated ± 5.6 pp, medium confidence
110GPT-5.260.9%estimated ± 5.3 pp, medium confidence
111Apodex 1.160.8%measured
112Apodex 1.1 Mini60.8%measured
113Kimi K2.7 Code60.8%measured
114MiMo-V2.5-Pro60.2%measured
115Claude Opus 4.659.4%estimated ± 4.4 pp, medium confidence
116Gemini 3 Flash59.3%estimated ± 5.6 pp, medium confidence
117Hy358.8%measured
118Hy3 Preview58.8%measured
119Muse Spark58.6%measured
120MiniMax M358.6%measured
121BTL-357.9%estimated ± 5.7 pp, medium confidence
122BTL-457.6%estimated ± 5.3 pp, medium confidence
123MiMo-V2-Pro56.8%estimated ± 5.3 pp, medium confidence
124GPT-5.4 mini56.1%measured
125GPT-5.4 nano56.1%measured
126Qwen3.7 Plus55.9%measured
127GPT-5.2-Codex55.8%estimated ± 5.6 pp, medium confidence
128GLM-5.155.8%measured
129Claude Sonnet 4.555.1%estimated ± 5.3 pp, medium confidence
130Qwen3.6 Plus54.5%measured
131Grok 4.2054.0%estimated ± 5.3 pp, medium confidence
132MiMo-V2.553.9%estimated ± 5.6 pp, medium confidence
133Qwen3.6-27B53.7%measured
134Inkling-Small52.9%measured
135MiniMax M2.752.6%measured
136Claude Sonnet 4.5 Thinking52.5%estimated ± 5.6 pp, medium confidence
137Inkling52.1%measured
138GLM-551.2%estimated ± 4.3 pp, high confidence
139GPT-5.1-Codex-Max51.0%estimated ± 6.9 pp, medium confidence
140Ling 3.0 Flash50.6%measured
141MiMo-V2-Omni49.9%estimated ± 5.3 pp, medium confidence
142MiMo-V2-Flash49.8%measured
143A.X K249.7%estimated ± 5.7 pp, medium confidence
144Laguna M.149.5%estimated ± 5.3 pp, medium confidence
145GPT-5.149.4%measured
146Gemini 3.5 Flash-Lite49.3%measured
147Claude 4.1 Opus49.3%estimated ± 5.3 pp, medium confidence
148Nemotron 3 Ultra49.3%measured
149Ornith-1.0-35B49.2%estimated ± 4.3 pp, high confidence
150Step 3.5 Flash49.1%estimated ± 5.3 pp, medium confidence
151Muse Glimmer 30B49.0%measured
152Flint48.6%estimated ± 5.3 pp, medium confidence
153Ling 3.0 Flash VL48.5%estimated ± 6.1 pp, medium confidence
154Qwen3.5 Plus47.6%estimated ± 6.9 pp, medium confidence
155MiniMax M2.547.2%estimated ± 6.9 pp, medium confidence
156MAI-Thinking-147.2%estimated ± 5.3 pp, medium confidence
157Mistral Medium 3.5 128B46.9%measured
158GPT-5 mini46.8%estimated ± 6.9 pp, medium confidence
159Kimi K2.546.8%measured
160Claude Haiku 4.546.7%estimated ± 5.3 pp, medium confidence
161GPT-5.1-Codex46.3%estimated ± 6.9 pp, medium confidence
162Ornith-1.5-9B46.1%estimated ± 4.3 pp, high confidence
163Qwen3.5 397B45.8%estimated ± 4.3 pp, high confidence
164GLM-4.545.7%estimated ± 11.5 pp, low confidence
165Jais-3-MoVA-36B-A4B45.7%estimated ± 5.3 pp, medium confidence
166Qwen3.5-122B-A10B45.7%measured
167Claude 4 Sonnet45.5%estimated ± 5.3 pp, medium confidence
168Claude Haiku 4.5 Thinking45.4%estimated ± 6.9 pp, medium confidence
169MAI-Code-1.1-Flash45.3%estimated ± 5.3 pp, medium confidence
170GLM-4.745.3%measured
171Claude 4 Opus45.1%estimated ± 5.3 pp, medium confidence
172Jais-3-32B44.9%estimated ± 5.3 pp, medium confidence
173Qwen3.5-27B44.9%estimated ± 5.3 pp, medium confidence
174Qwen3.5 Flash44.1%estimated ± 5.6 pp, medium confidence
175Gemma 4 31B43.4%measured
176Nemotron 3.5 Super VL 120B A12B42.3%estimated ± 5.3 pp, medium confidence
177Nemotron 3.5 Super VL 120B A12B NVFP442.3%estimated ± 5.3 pp, medium confidence
178Grok 4.342.3%measured
179Laguna XS 2.142.0%estimated ± 5.3 pp, medium confidence
180Qwen3.6-35B-A3B41.9%measured
181Grok Code Fast 141.8%estimated ± 5.3 pp, medium confidence
182Gemini 3.1 Flash-Lite41.7%estimated ± 5.6 pp, medium confidence
183Solar Pro 441.4%estimated ± 5.3 pp, medium confidence
184Qwen3 Max41.3%estimated ± 6.9 pp, medium confidence
185GLM-4.641.1%estimated ± 6.9 pp, medium confidence
186Solar Open 241.0%estimated ± 5.3 pp, medium confidence
187Ling 3.0 Flash FP840.8%estimated ± 9.2 pp, medium confidence
188Jais-3-375B-A23B40.7%estimated ± 5.3 pp, medium confidence
189Jais-3-7B40.7%estimated ± 5.3 pp, medium confidence
190Ornith-1.0-9B40.5%estimated ± 4.3 pp, medium confidence
191Laguna XS.240.1%estimated ± 5.3 pp, medium confidence
192DeepSeek V3.240.1%estimated ± 11.1 pp, low confidence
193o139.7%measured
194Qwen3-Coder 480B-A35B39.6%estimated ± 5.3 pp, medium confidence
195Step 3.7 Flash39.6%measured
196Gemma 4 26B A4B39.3%measured
197Qwen3.5-35B-A3B38.9%estimated ± 5.3 pp, medium confidence
198Claude 3.7 Sonnet Thinking38.4%estimated ± 11.5 pp, low confidence
199GPT-537.8%measured
200K-EXAONE 2.037.3%estimated ± 5.3 pp, medium confidence
201LongCat-Flash-Lite-Sparse37.3%estimated ± 5.3 pp, medium confidence
202Nemotron 3.5 Bolt 74B-A9B36.7%estimated ± 6.1 pp, medium confidence
203DeepSeek V3 032435.0%estimated ± 6.1 pp, medium confidence
204Grok 434.7%estimated ± 5.6 pp, medium confidence
205North Mini Code34.5%estimated ± 6.1 pp, medium confidence
206DeepSeek V3.134.1%estimated ± 5.3 pp, medium confidence
207o1-preview34.1%measured
208Kimi K233.8%estimated ± 5.3 pp, medium confidence
209Mercury 2.533.8%estimated ± 6.1 pp, medium confidence
210Gemini 2.5 Pro33.3%measured
211K-Exaone32.1%measured
212Gemma 4 12B31.0%measured
213Spark-X2.6-4B30.5%estimated ± 5.3 pp, medium confidence
214GPT-OSS 120B30.4%measured
215Claude 3.7 Sonnet30.0%estimated ± 5.3 pp, medium confidence
216Ternary Bonsai 2 27B28.8%estimated ± 5.3 pp, medium confidence
217Nemotron 3 Super 100B28.6%estimated ± 5.3 pp, medium confidence
218Command A+27.9%measured
219GLM-4.7-Flash27.7%estimated ± 5.3 pp, medium confidence
220DeepSeek-R1-052826.8%estimated ± 5.3 pp, medium confidence
221Nemotron 3.5 Lightning 30B A3B NVFP426.8%measured
222Mistral Small 426.6%measured
223Granite 4.2 30B26.5%estimated ± 5.3 pp, medium confidence
224MiniMax M1 80k26.1%estimated ± 5.3 pp, medium confidence
225Trinity-Large-Preview25.8%measured
226Trinity-Large-Thinking25.8%measured
227GPT-4.125.6%estimated ± 5.3 pp, medium confidence
228Ling 2.6 Flash25.3%measured
229ZAYA1-74B-Preview25.2%estimated ± 5.3 pp, medium confidence
230LFM2.5-24B-A2B24.7%estimated ± 5.3 pp, medium confidence
231o3-mini24.4%estimated ± 5.3 pp, medium confidence
232LLaDA2.2-flash24.4%estimated ± 5.3 pp, medium confidence
233DeepSeek-R124.3%estimated ± 5.3 pp, medium confidence
234Claude 3.5 Sonnet24.3%estimated ± 5.3 pp, medium confidence
235Mellum2.1-12B-A2.5B-Thinking24.1%estimated ± 5.3 pp, medium confidence
236Devstral Small24.0%estimated ± 5.3 pp, medium confidence
237MiniCPM5-2B24.0%estimated ± 5.3 pp, medium confidence
238Mellum2-12B-A2.5B-Thinking23.9%estimated ± 5.7 pp, medium confidence
239Sarvam 105B23.9%estimated ± 5.3 pp, medium confidence
240LFM3-1.7B23.7%estimated ± 5.3 pp, medium confidence
241Granite 4.2 3B23.7%estimated ± 5.7 pp, medium confidence
242Gemini 1.5 Pro23.6%measured
243Claude 3.5 Haiku23.6%estimated ± 5.3 pp, medium confidence
244Sarvam 30B23.4%estimated ± 5.3 pp, medium confidence
245Gemini 2.5 Flash-Lite23.4%estimated ± 5.3 pp, medium confidence
246Gemini 2.0 Flash23.3%estimated ± 5.3 pp, low confidence
247Gemini 2.0 Flash-Lite23.3%estimated ± 5.3 pp, low confidence
248Gemini 1.5 Flash23.3%estimated ± 5.3 pp, low confidence
249DeepSeek V323.0%measured
250Grok 4 Fast (Reasoning)22.6%estimated ± 5.6 pp, medium confidence
251Granite 4.2 8B22.4%measured
252GPT-4 Turbo21.5%measured
253GPT-OSS 20B20.7%measured
254Grok 4.1 Fast (Reasoning)20.6%estimated ± 5.6 pp, medium confidence
255GPT-4o20.5%estimated ± 11.5 pp, low confidence
256GPT-4.1 mini20.2%measured
257Mistral Large 320.1%measured
258Claude 3 Opus19.5%measured
259ZAYA1-8B18.3%estimated ± 5.7 pp, medium confidence
260Llama 4 Maverick16.3%measured
261Mistral Small 3.115.2%estimated ± 6.1 pp, medium confidence
262Celeris-114.4%measured
263Nemotron 3 Nano 30B14.4%measured
264Nemotron 3 Nano Omni 30B A3B13.8%measured
265Solar Pro 313.0%estimated ± 6.1 pp, medium confidence
266Ling 3.0 Tiny12.0%estimated ± 6.1 pp, medium confidence
267Ultravox v0.6 Llama 3.3 70B11.9%measured
268Agents-A1-4B11.8%estimated ± 5.7 pp, medium confidence
269GPT-4o mini11.4%measured
270GPT-4.1 nano11.1%measured
271Gemma 3 27B10.1%measured
272Gemma 4 E4B9.4%measured
273Llama 4 Scout8.2%measured
274LFM2.5-2.6B7.7%measured
275Gemma 4 E2B7.2%measured
276Qwen3 235B 25076.7%estimated ± 5.7 pp, medium confidence
277Magistral Medium6.1%estimated ± 5.7 pp, medium confidence
278Magistral Small5.0%estimated ± 5.7 pp, medium confidence
279Mellum2-12B-A2.5B-Instruct3.1%estimated ± 5.7 pp, low confidence
280MiniCPM5-1B2.9%estimated ± 5.7 pp, low confidence
281Mistral Medium 32.7%estimated ± 5.7 pp, low confidence
282LLaDA2.2-mini2.7%estimated ± 5.7 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General