Coding

AA Coding Index leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on AA Coding Index is 81.6% by Claude Fable 5.1. 117 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Composer 2	88.9%	estimated ± 11.4 pp, low confidence
2	Composer 2 Fast	87.1%	estimated ± 11.4 pp, low confidence
3	Claude Opus 5.5	83.9%	estimated ± 1.0 pp, medium confidence
4	Claude Sonnet 5.5	83.0%	estimated ± 1.0 pp, medium confidence
5	Claude Fable 5.1	81.6%	measured
6	Grok 4.7	79.0%	estimated ± 1.0 pp, high confidence
7	Claude Mythos 5	78.9%	estimated ± 5.1 pp, medium confidence
8	Gemini 4 Argon	78.6%	estimated ± 2.5 pp, medium confidence
9	Claude Opus 5	78.0%	measured
10	Ember-1	77.8%	estimated ± 2.5 pp, high confidence
11	DeepSeek V4.1 Flash	77.5%	estimated ± 2.5 pp, high confidence
12	Pareto 26.9	77.4%	estimated ± 2.5 pp, high confidence
13	GPT-5.6 Sol	77.4%	measured
14	SWE-2	77.1%	estimated ± 2.5 pp, high confidence
15	GPT-6 Astra	76.9%	measured
16	Grok 4.6	76.8%	measured
17	GPT-6.1 Sol	76.7%	estimated ± 2.5 pp, high confidence
18	MiMo-V2.6-Pro	76.7%	estimated ± 2.5 pp, high confidence
19	GPT-5.6 Terra	76.7%	measured
20	Claude Fable 5	76.5%	measured
21	Gemini 3.8 Flash	76.3%	measured
22	Kimi K3	76.2%	measured
23	Pareto 26.10 Preview	76.1%	estimated ± 2.5 pp, high confidence
24	Gemini 3.7 Flash	76.1%	measured
25	GPT-6 Sol	75.8%	estimated ± 2.5 pp, high confidence
26	Muse Spark 1.3	75.8%	measured
27	MiMo-V2.6-Flash	75.5%	estimated ± 2.5 pp, high confidence
28	Step 5 Preview	75.5%	estimated ± 2.5 pp, high confidence
29	Claude Haiku 5.5	75.2%	estimated ± 3.3 pp, high confidence
30	GPT-6 Luna	75.1%	estimated ± 2.5 pp, high confidence
31	GPT-5.5	74.9%	measured
32	GLM-5.3	74.8%	measured
33	Hy4 preview	74.4%	estimated ± 2.5 pp, high confidence
34	Claude Opus 4.8	74.3%	measured
35	GLM-5.3-Flash	74.1%	estimated ± 2.5 pp, high confidence
36	Claude Opus 4.7 (Adaptive)	73.6%	measured
37	Claude Opus 4.7	73.5%	estimated ± 3.1 pp, high confidence
38	SWE-1.7	73.5%	estimated ± 3.3 pp, high confidence
39	Qwen3.8-Flash-Next	73.1%	measured
40	Grok 4.5	72.5%	measured
41	Qwen3.8-Omni-Flash	72.4%	estimated ± 2.5 pp, high confidence
42	Muse Spark 1.2	72.2%	measured
43	Qwen3.8 Max	72.0%	estimated ± 2.5 pp, high confidence
44	Ornith-1.5-397B	71.8%	estimated ± 2.5 pp, high confidence
45	Qwen3.8 Max Preview	71.8%	measured
46	Claude Sonnet 5	71.6%	measured
47	GPT-5.6 Luna	71.5%	measured
48	Muse Spark 1.1	71.3%	measured
49	GPT-5.4	71.1%	measured
50	Mistral Large 4	70.6%	estimated ± 6.4 pp, medium confidence
51	Ling 3.1 Flash	70.4%	estimated ± 6.4 pp, medium confidence
52	Gemini 3.5 Flash	70.1%	measured
53	Gemini 3.6 Flash	69.2%	measured
54	DeepSeek V4 Flash 0731	69.1%	measured
55	DeepSeek V4 Pro 0813	68.8%	measured
56	Gemini 3.1 Pro	68.8%	measured
57	GLM-5.2	68.8%	measured
58	GPT-5.3 Codex	68.6%	estimated ± 5.1 pp, medium confidence
59	Beam	68.3%	estimated ± 2.5 pp, high confidence
60	dots3-note Preview	68.1%	estimated ± 4.6 pp, high confidence
61	Qwen3.8-27B	68.1%	measured
62	Claude Opus 4.6 (Adaptive)	67.8%	estimated ± 6.8 pp, medium confidence
63	Sakana Fugu-Ultra	67.5%	estimated ± 6.4 pp, low confidence
64	Laguna S 2.1	67.0%	estimated ± 2.5 pp, medium confidence
65	Ornith-1.0-397B	66.9%	estimated ± 4.6 pp, high confidence
66	Sakana Fugu	66.9%	estimated ± 6.4 pp, low confidence
67	Qwen3.7 Max	66.0%	measured
68	Composer 2.5	65.9%	estimated ± 1.0 pp, medium confidence
69	Seed 2.1 Pro	65.8%	estimated ± 4.6 pp, high confidence
70	Atria Dawn Preview	64.8%	estimated ± 7.5 pp, medium confidence
71	Seed 2.1 Turbo	62.3%	estimated ± 4.6 pp, high confidence
72	Kimi K2.6	61.8%	measured
73	Claude Opus 4.5	61.7%	estimated ± 4.6 pp, high confidence
74	Ornith-1.5-35B-A3B	61.4%	estimated ± 2.5 pp, medium confidence
75	Qwen 3.6 Max (preview)	61.3%	estimated ± 4.6 pp, high confidence
76	Quasar 438B	61.2%	measured
77	Apodex 1.1	60.8%	measured
78	Apodex 1.1 Mini	60.8%	measured
79	Kimi K2.7 Code	60.8%	measured
80	GPT-5.2	60.2%	estimated ± 5.1 pp, medium confidence
81	MiMo-V2.5-Pro	60.2%	measured
82	GLM-4.5	59.6%	estimated ± 9.0 pp, low confidence
83	Gemini 3 Flash	59.4%	estimated ± 5.8 pp, medium confidence
84	Hy3	58.8%	measured
85	Hy3 Preview	58.8%	measured
86	Muse Spark	58.6%	measured
87	MiniMax M3	58.6%	measured
88	BTL-3	57.4%	estimated ± 6.4 pp, medium confidence
89	BTL-4	57.2%	estimated ± 5.1 pp, medium confidence
90	MiMo-V2-Pro	56.4%	estimated ± 5.1 pp, medium confidence
91	GPT-5.2-Codex	56.2%	estimated ± 5.8 pp, medium confidence
92	GPT-5.4 mini	56.1%	measured
93	GPT-5.4 nano	56.1%	measured
94	GLM-5	56.0%	estimated ± 5.1 pp, medium confidence
95	Claude Opus 4.6	55.9%	estimated ± 3.3 pp, medium confidence
96	Qwen3.7 Plus	55.9%	measured
97	GLM-5.1	55.8%	measured
98	Claude Sonnet 4.5	54.8%	estimated ± 5.1 pp, medium confidence
99	Qwen3.6 Plus	54.5%	measured
100	MiMo-V2.5	54.3%	estimated ± 5.8 pp, medium confidence
101	Grok 4.20	53.8%	estimated ± 5.1 pp, medium confidence
102	Qwen3.6-27B	53.7%	measured
103	Inkling-Small	52.9%	measured
104	Qwen3.5 397B	52.8%	estimated ± 5.1 pp, medium confidence
105	MiniMax M2.7	52.6%	measured
106	Inkling	52.1%	measured
107	GLM-5 (Reasoning)	51.4%	estimated ± 6.8 pp, medium confidence
108	Claude Sonnet 4.5 Thinking	51.0%	estimated ± 6.8 pp, medium confidence
109	GPT-5.1-Codex-Max	50.7%	estimated ± 6.8 pp, medium confidence
110	Ling 3.0 Flash	50.6%	measured
111	Ling 3.0 Flash FP8	50.6%	measured
112	MiMo-V2-Omni	50.0%	estimated ± 5.1 pp, medium confidence
113	Claude Opus 4.5 Thinking	49.9%	estimated ± 6.8 pp, medium confidence
114	MiMo-V2-Flash	49.8%	measured
115	Laguna M.1	49.6%	estimated ± 5.1 pp, medium confidence
116	Claude Sonnet 4.6	49.5%	estimated ± 3.3 pp, medium confidence
117	Claude 4.1 Opus	49.4%	estimated ± 5.1 pp, medium confidence
118	GPT-5.1	49.4%	measured
119	Gemini 3.5 Flash-Lite	49.3%	measured
120	Nemotron 3 Ultra	49.3%	measured
121	Muse Glimmer 30B	49.0%	measured
122	Ornith-1.0-35B	48.8%	estimated ± 4.6 pp, high confidence
123	Ling 3.0 Flash VL	48.7%	estimated ± 6.4 pp, medium confidence
124	MAI-Thinking-1	47.5%	estimated ± 5.1 pp, medium confidence
125	Qwen3.5 Plus	47.2%	estimated ± 6.8 pp, medium confidence
126	Claude Haiku 4.5	47.1%	estimated ± 5.1 pp, medium confidence
127	Mistral Medium 3.5 128B	46.9%	measured
128	Kimi K2.5	46.8%	measured
129	Kimi K2.5 (Reasoning)	46.8%	measured
130	MiniMax M2.5	46.7%	estimated ± 6.8 pp, medium confidence
131	Gemini 3 Pro	46.4%	estimated ± 6.8 pp, medium confidence
132	GPT-5 mini	46.4%	estimated ± 6.8 pp, medium confidence
133	Claude 4 Sonnet	46.0%	estimated ± 5.1 pp, medium confidence
134	GPT-5.1-Codex	45.8%	estimated ± 6.8 pp, medium confidence
135	MAI-Code-1.1-Flash	45.8%	estimated ± 5.1 pp, medium confidence
136	Qwen3.5-122B-A10B	45.7%	measured
137	Ornith-1.5-9B	45.5%	estimated ± 4.6 pp, high confidence
138	Qwen3.5-27B	45.4%	estimated ± 5.1 pp, medium confidence
139	Qwen3.5 Flash	45.3%	estimated ± 5.8 pp, medium confidence
140	GLM-4.7	45.3%	measured
141	Claude Haiku 4.5 Thinking	44.9%	estimated ± 6.8 pp, medium confidence
142	Gemma 4 31B	43.4%	measured
143	Gemini 3.1 Flash-Lite	43.1%	estimated ± 5.8 pp, medium confidence
144	Laguna XS 2.1	42.7%	estimated ± 5.1 pp, medium confidence
145	Grok Code Fast 1	42.5%	estimated ± 5.1 pp, medium confidence
146	Grok 4.3	42.3%	measured
147	Solar Pro 4	42.1%	estimated ± 5.1 pp, medium confidence
148	Qwen3.6-35B-A3B	41.9%	measured
149	Solar Open 2	41.8%	estimated ± 5.1 pp, medium confidence
150	Grok 4	41.6%	estimated ± 11.4 pp, low confidence
151	DeepSeek V3.2 (Thinking)	41.4%	estimated ± 6.8 pp, medium confidence
152	Laguna XS.2	40.9%	estimated ± 5.1 pp, medium confidence
153	Qwen3 Max	40.5%	estimated ± 6.8 pp, medium confidence
154	GLM-4.6	40.3%	estimated ± 6.8 pp, medium confidence
155	A.X K2	40.3%	estimated ± 6.4 pp, medium confidence
156	Qwen3.5-35B-A3B	39.7%	estimated ± 5.1 pp, medium confidence
157	o1	39.7%	measured
158	Ornith-1.0-9B	39.6%	estimated ± 4.6 pp, medium confidence
159	Step 3.7 Flash	39.6%	measured
160	Gemma 4 26B A4B	39.3%	measured
161	Grok 4.1 Fast (Reasoning)	39.3%	estimated ± 6.8 pp, medium confidence
162	DeepSeek V3.2	39.1%	estimated ± 11.4 pp, low confidence
163	Grok 4 Fast (Reasoning)	38.6%	estimated ± 6.8 pp, low confidence
164	K-EXAONE 2.0	38.2%	estimated ± 5.1 pp, medium confidence
165	LongCat-Flash-Lite-Sparse	38.2%	estimated ± 5.1 pp, medium confidence
166	GPT-5 (high)	37.8%	measured
167	Nemotron 3 Super 100B	37.7%	measured
168	DeepSeek V3 0324	35.2%	estimated ± 6.4 pp, medium confidence
169	North Mini Code	34.7%	estimated ± 6.4 pp, medium confidence
170	o1-preview	34.1%	measured
171	Mercury 2.5	34.0%	estimated ± 6.4 pp, medium confidence
172	Gemini 2.5 Pro	33.3%	measured
173	K-Exaone	32.1%	measured
174	Gemma 4 12B	31.0%	measured
175	GPT-OSS 120B	30.4%	measured
176	Ternary Bonsai 2 27B	29.4%	estimated ± 5.1 pp, medium confidence
177	Command A+	27.9%	measured
178	Nemotron 3.5 Lightning 30B A3B NVFP4	26.8%	measured
179	Granite 4.2 30B	26.7%	estimated ± 5.1 pp, medium confidence
180	Mistral Small 4	26.6%	measured
181	Mistral Small 4 (Reasoning)	26.6%	measured
182	Trinity-Large-Preview	25.8%	measured
183	Trinity-Large-Thinking	25.8%	measured
184	GPT-4.1	25.5%	estimated ± 5.1 pp, medium confidence
185	Ling 2.6 Flash	25.3%	measured
186	ZAYA1-74B-Preview	24.9%	estimated ± 5.1 pp, medium confidence
187	o3-mini	23.8%	estimated ± 5.1 pp, medium confidence
188	LLaDA2.2-flash	23.8%	estimated ± 5.1 pp, medium confidence
189	Claude 3.5 Sonnet	23.7%	estimated ± 5.1 pp, medium confidence
190	Gemini 1.5 Pro	23.6%	measured
191	MiniCPM5-2B	23.2%	estimated ± 5.1 pp, medium confidence
192	DeepSeek V3	23.0%	measured
193	Granite 4.2 8B	22.4%	measured
194	Mellum2-12B-A2.5B-Thinking	22.3%	estimated ± 6.4 pp, medium confidence
195	GPT-4 Turbo	21.5%	measured
196	GPT-OSS 20B	20.7%	measured
197	GPT-4.1 mini	20.2%	measured
198	Mistral Large 3	20.1%	measured
199	Claude 3 Opus	19.5%	measured
200	ZAYA1-8B	16.5%	estimated ± 6.4 pp, medium confidence
201	Llama 4 Maverick	16.3%	measured
202	Celeris-1	14.4%	measured
203	Nemotron 3 Nano 30B	14.4%	measured
204	Nemotron 3 Nano Omni 30B A3B	13.8%	measured
205	Solar Pro 3	12.9%	estimated ± 6.4 pp, medium confidence
206	Granite 4.2 3B	12.7%	estimated ± 6.4 pp, medium confidence
207	Ultravox v0.6 Llama 3.3 70B	11.9%	measured
208	Ling 3.0 Tiny	11.8%	estimated ± 6.4 pp, medium confidence
209	GPT-4o mini	11.4%	measured
210	GPT-4.1 nano	11.1%	measured
211	Gemma 3 27B	10.1%	measured
212	Agents-A1-4B	9.7%	estimated ± 6.4 pp, medium confidence
213	Gemma 4 E4B	9.4%	measured
214	Llama 4 Scout	8.2%	measured
215	LFM2.5-2.6B	7.7%	measured
216	Gemma 4 E2B	7.2%	measured
217	Mellum2-12B-A2.5B-Instruct	0.6%	estimated ± 6.4 pp, low confidence
218	MiniCPM5-1B	0.3%	estimated ± 6.4 pp, low confidence
219	LLaDA2.2-mini	0.1%	estimated ± 6.4 pp, low confidence

AA Coding Index leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General