Coding

SciCode leaderboard

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

As of 2026-10-07, the highest measured score on SciCode is 60.1% by Sakana Fugu. 168 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Opus 5.5	66.1%	estimated ± 3.2 pp, medium confidence
2	Gemini 4 Argon	61.1%	estimated ± 3.2 pp, medium confidence
3	Claude Sonnet 5.5	60.3%	estimated ± 3.2 pp, medium confidence
4	MiMo-V2.6-Pro	60.3%	estimated ± 3.2 pp, medium confidence
5	Sakana Fugu	60.1%	measured
6	Step 5 Preview	58.9%	measured
7	Sakana Fugu-Ultra	58.7%	measured
8	Muse Spark 1.3	58.2%	estimated ± 3.2 pp, high confidence
9	Claude Fable 5.1	57.8%	estimated ± 2.9 pp, low confidence
10	GPT-6 Sol	57.1%	estimated ± 3.2 pp, high confidence
11	Grok 4.7	56.9%	estimated ± 3.2 pp, high confidence
12	Muse Spark 1.2	56.9%	estimated ± 3.2 pp, high confidence
13	Claude Fable 5	56.4%	estimated ± 2.9 pp, low confidence
14	GPT-6 Astra	56.0%	estimated ± 3.2 pp, high confidence
15	Gemini 3.8 Flash	55.8%	estimated ± 2.9 pp, low confidence
16	Claude Mythos 5	55.3%	estimated ± 3.8 pp, medium confidence
17	Claude Opus 5	54.8%	estimated ± 2.9 pp, low confidence
18	Claude Haiku 5.5	54.5%	estimated ± 3.2 pp, high confidence
19	Gemini 3.7 Flash	54.2%	estimated ± 2.9 pp, low confidence
20	GPT-6 Luna	54.2%	estimated ± 3.2 pp, high confidence
21	Hy4 preview	54.1%	estimated ± 4.8 pp, high confidence
22	GPT-5.4	54.0%	estimated ± 4.7 pp, medium confidence
23	Ember-1	53.9%	estimated ± 3.8 pp, medium confidence
24	GPT-6.1 Sol	53.8%	estimated ± 3.2 pp, high confidence
25	Mistral Large 4	53.8%	estimated ± 3.2 pp, high confidence
26	Gemini 3.1 Pro	53.8%	estimated ± 2.9 pp, low confidence
27	Ling 3.1 Flash	53.7%	estimated ± 3.2 pp, high confidence
28	Qwen3.7 Max	53.5%	measured
29	GPT-5.6 Luna	53.2%	estimated ± 3.2 pp, high confidence
30	Grok 4.6	53.1%	estimated ± 2.9 pp, low confidence
31	Gemini 3.5 Flash	53.1%	measured
32	Qwen3.8-Omni-Flash	53.0%	estimated ± 4.8 pp, high confidence
33	Gemini 3.6 Flash	52.9%	estimated ± 2.9 pp, low confidence
34	GPT-5.2-Codex	52.7%	estimated ± 2.9 pp, low confidence
35	Qwen3.8 Max	52.5%	estimated ± 2.9 pp, low confidence
36	Claude Opus 4.8	52.3%	estimated ± 2.9 pp, low confidence
37	Kimi K2.6	52.2%	measured
38	Claude Opus 4.7 (Adaptive)	51.9%	estimated ± 3.8 pp, medium confidence
39	DeepSeek V4 Pro 0813	51.7%	estimated ± 2.9 pp, medium confidence
40	Qwen3.8 Max Preview	51.7%	estimated ± 3.2 pp, high confidence
41	DeepSeek V4.1 Flash	51.5%	estimated ± 3.2 pp, high confidence
42	Grok 4.5	51.5%	estimated ± 2.9 pp, medium confidence
43	DeepSeek V4 Flash 0731	51.3%	estimated ± 2.9 pp, medium confidence
44	GPT-5.3 Codex	51.3%	estimated ± 2.9 pp, medium confidence
45	Qwen3.7 Plus	51.3%	measured
46	Atria Dawn Preview	51.2%	estimated ± 4.8 pp, high confidence
47	Ornith-1.5-397B	51.2%	estimated ± 3.8 pp, medium confidence
48	Laguna S 2.1	51.1%	estimated ± 4.8 pp, high confidence
49	Composer 2.5	51.1%	estimated ± 4.5 pp, medium confidence
50	Kimi K3	51.1%	estimated ± 2.9 pp, medium confidence
51	MiMo-V2.6-Flash	51.0%	estimated ± 3.2 pp, high confidence
52	Qwen3.8-Flash-Next	50.3%	estimated ± 3.2 pp, high confidence
53	BTL-3	49.8%	estimated ± 5.4 pp, medium confidence
54	Beam	49.7%	measured
55	Ornith-1.0-397B	49.6%	estimated ± 3.8 pp, medium confidence
56	Claude Opus 4.5	48.9%	estimated ± 3.8 pp, high confidence
57	Claude Opus 4.6	48.9%	estimated ± 3.8 pp, high confidence
58	Qwen3.6 Plus	48.7%	estimated ± 2.9 pp, medium confidence
59	Inkling-Small	48.7%	measured
60	Kimi K2.5	48.7%	measured
61	GPT-5.2	48.5%	estimated ± 3.8 pp, high confidence
62	GPT-5.6 Terra	48.5%	estimated ± 2.9 pp, medium confidence
63	Muse Spark 1.1	48.5%	estimated ± 2.9 pp, medium confidence
64	Hy3	48.3%	estimated ± 3.2 pp, high confidence
65	Ornith-1.5-35B-A3B	48.1%	estimated ± 3.8 pp, high confidence
66	Gemini 3 Flash	47.9%	estimated ± 2.9 pp, medium confidence
67	GPT-5.1-Codex	47.9%	estimated ± 2.9 pp, medium confidence
68	Quasar 438B	47.9%	estimated ± 3.2 pp, high confidence
69	BTL-4	47.8%	estimated ± 3.8 pp, high confidence
70	dots3-note Preview	47.8%	estimated ± 3.8 pp, high confidence
71	Inkling	47.7%	estimated ± 2.9 pp, medium confidence
72	MiMo-V2-Pro	47.6%	estimated ± 3.8 pp, high confidence
73	GLM-5	47.5%	estimated ± 3.8 pp, high confidence
74	Muse Spark	47.3%	estimated ± 3.8 pp, high confidence
75	Grok 4.3	47.3%	measured
76	GPT-5.5	47.3%	estimated ± 2.9 pp, medium confidence
77	Claude Sonnet 4.5	47.2%	estimated ± 3.8 pp, high confidence
78	Kimi K2.5 (Reasoning)	47.0%	estimated ± 3.8 pp, high confidence
79	Qwen 3.6 Max (preview)	47.0%	measured
80	Claude Opus 4.7	46.9%	estimated ± 2.9 pp, medium confidence
81	Qwen3.5 397B	46.7%	estimated ± 3.8 pp, high confidence
82	Ornith-1.0-35B	46.5%	estimated ± 3.8 pp, high confidence
83	Gemini 2.5 Pro	46.1%	estimated ± 3.2 pp, high confidence
84	MiMo-V2-Omni	46.1%	estimated ± 3.8 pp, high confidence
85	Claude 4.1 Opus	45.9%	estimated ± 3.8 pp, high confidence
86	MAI-Thinking-1	45.5%	estimated ± 3.8 pp, high confidence
87	MiMo-V2-Flash	45.4%	estimated ± 3.8 pp, high confidence
88	Apodex 1.1	45.3%	estimated ± 3.2 pp, high confidence
89	Apodex 1.1 Mini	45.3%	estimated ± 3.2 pp, high confidence
90	Gemma 4 31B	45.3%	estimated ± 3.2 pp, high confidence
91	Grok 4.20	45.3%	estimated ± 2.9 pp, medium confidence
92	Claude 4 Sonnet	45.1%	estimated ± 3.8 pp, high confidence
93	MAI-Code-1.1-Flash	45.0%	estimated ± 3.8 pp, high confidence
94	Qwen3.5-27B	44.9%	estimated ± 3.8 pp, high confidence
95	GPT-5.4 nano	44.7%	estimated ± 2.9 pp, medium confidence
96	Qwen3.8-27B	44.7%	estimated ± 2.9 pp, medium confidence
97	Nemotron 3 Ultra	44.6%	measured
98	GPT-5.1	44.6%	estimated ± 4.7 pp, high confidence
99	Solar Pro 4	44.5%	estimated ± 3.2 pp, high confidence
100	Laguna XS 2.1	44.2%	estimated ± 3.8 pp, high confidence
101	Ling 3.0 Flash VL	44.1%	estimated ± 3.2 pp, high confidence
102	Ornith-1.5-9B	44.1%	estimated ± 3.8 pp, high confidence
103	Solar Open 2	44.0%	estimated ± 3.8 pp, high confidence
104	GPT-5.1-Codex-Max	43.9%	estimated ± 2.9 pp, low confidence
105	Step 3.7 Flash	43.8%	estimated ± 3.2 pp, high confidence
106	Muse Glimmer 30B	43.6%	measured
107	Ornith-1.0-9B	43.5%	estimated ± 3.8 pp, high confidence
108	Qwen3.5-35B-A3B	43.4%	estimated ± 3.8 pp, high confidence
109	Qwen3.5 Flash	43.3%	estimated ± 2.9 pp, low confidence
110	LongCat-Flash-Lite-Sparse	42.9%	estimated ± 3.8 pp, high confidence
111	Qwen3.6-27B	42.7%	estimated ± 3.2 pp, high confidence
112	GPT-5.6 Sol	41.9%	estimated ± 2.9 pp, low confidence
113	Claude Sonnet 5	41.5%	estimated ± 2.9 pp, low confidence
114	Ling 3.0 Flash	41.2%	measured
115	Hy3 Preview	41.2%	measured
116	GLM-4.7	41.1%	estimated ± 2.9 pp, low confidence
117	MiniMax M3	41.1%	estimated ± 2.9 pp, low confidence
118	A.X K2	41.0%	measured
119	Claude Sonnet 4.6	40.9%	estimated ± 2.9 pp, low confidence
120	Kimi K2.7 Code	40.9%	estimated ± 2.9 pp, low confidence
121	Trinity-Large-Preview	40.6%	estimated ± 3.2 pp, high confidence
122	Trinity-Large-Thinking	40.6%	estimated ± 3.2 pp, high confidence
123	Ling 3.0 Flash FP8	40.4%	measured
124	o1	40.4%	estimated ± 4.7 pp, high confidence
125	Mistral Medium 3.5 128B	40.2%	estimated ± 3.2 pp, high confidence
126	Gemma 4 26B A4B	40.0%	estimated ± 3.2 pp, high confidence
127	Qwen3.5-122B-A10B	39.7%	estimated ± 3.2 pp, high confidence
128	GPT-5.4 mini	39.7%	estimated ± 2.9 pp, low confidence
129	MiMo-V2.5	39.7%	estimated ± 2.9 pp, low confidence
130	GLM-5.1	39.5%	estimated ± 2.9 pp, low confidence
131	MiMo-V2.5-Pro	39.5%	estimated ± 2.9 pp, low confidence
132	GPT-5 (high)	39.5%	estimated ± 4.7 pp, high confidence
133	Ternary Bonsai 2 27B	39.1%	estimated ± 3.8 pp, high confidence
134	DeepSeek V3 0324	39.1%	estimated ± 3.2 pp, high confidence
135	GPT-OSS 20B	39.0%	estimated ± 3.2 pp, high confidence
136	Mistral Small 4	38.9%	estimated ± 3.2 pp, high confidence
137	Mistral Small 4 (Reasoning)	38.9%	estimated ± 3.2 pp, high confidence
138	North Mini Code	38.9%	estimated ± 3.2 pp, high confidence
139	Granite 4.2 30B	38.8%	measured
140	GLM-4.6	38.8%	estimated ± 2.9 pp, low confidence
141	Command A+	38.6%	estimated ± 3.2 pp, high confidence
142	Mercury 2.5	38.0%	measured
143	o1-preview	37.9%	estimated ± 4.7 pp, high confidence
144	GLM-5.3	37.8%	estimated ± 2.9 pp, low confidence
145	GLM-5.3-Flash	37.8%	estimated ± 2.9 pp, low confidence
146	K-EXAONE 2.0	37.4%	measured
147	K-Exaone	37.1%	estimated ± 4.7 pp, high confidence
148	Gemini 3.1 Flash-Lite	37.0%	estimated ± 2.9 pp, low confidence
149	Mistral Large 3	36.7%	estimated ± 3.2 pp, high confidence
150	Qwen3.6-35B-A3B	36.7%	estimated ± 3.2 pp, high confidence
151	MiniMax M2.7	36.6%	estimated ± 2.9 pp, low confidence
152	Gemma 4 12B	36.6%	estimated ± 4.7 pp, high confidence
153	Nemotron 3 Super 100B	36.3%	estimated ± 3.2 pp, high confidence
154	Granite 4.2 8B	36.1%	measured
155	DeepSeek V3	36.0%	estimated ± 3.2 pp, high confidence
156	GPT-4.1	35.9%	estimated ± 3.8 pp, high confidence
157	ZAYA1-74B-Preview	35.1%	estimated ± 3.8 pp, high confidence
158	Gemini 3.5 Flash-Lite	34.9%	estimated ± 2.9 pp, low confidence
159	GPT-OSS 120B	34.2%	estimated ± 3.2 pp, high confidence
160	Gemini 1.5 Pro	33.4%	estimated ± 4.7 pp, high confidence
161	o3-mini	33.0%	estimated ± 3.8 pp, high confidence
162	LLaDA2.2-flash	33.0%	estimated ± 3.8 pp, high confidence
163	Claude 3.5 Sonnet	32.8%	estimated ± 3.8 pp, high confidence
164	GPT-4 Turbo	32.4%	estimated ± 4.7 pp, high confidence
165	Nemotron 3 Nano Omni 30B A3B	32.0%	measured
166	Llama 4 Maverick	32.0%	estimated ± 3.2 pp, high confidence
167	Claude 3 Opus	31.6%	estimated ± 4.7 pp, high confidence
168	Nemotron 3.5 Lightning 30B A3B NVFP4	31.4%	measured
169	Nemotron 3 Nano 30B	30.9%	estimated ± 3.2 pp, high confidence
170	Mellum2-12B-A2.5B-Thinking	30.4%	estimated ± 5.4 pp, medium confidence
171	Agents-A1-4B	29.6%	measured
172	Ultravox v0.6 Llama 3.3 70B	28.3%	estimated ± 4.7 pp, medium confidence
173	GPT-4o mini	28.1%	estimated ± 4.7 pp, medium confidence
174	GPT-4.1 nano	28.0%	estimated ± 4.7 pp, medium confidence
175	ZAYA1-8B	27.2%	estimated ± 5.4 pp, medium confidence
176	Ling 2.6 Flash	27.0%	measured
177	MiniCPM5-2B	26.3%	measured
178	Gemma 4 E2B	26.3%	estimated ± 4.7 pp, medium confidence
179	Solar Pro 3	26.0%	estimated ± 3.2 pp, high confidence
180	Gemma 4 E4B	24.9%	estimated ± 3.2 pp, medium confidence
181	Ling 3.0 Tiny	24.7%	estimated ± 3.2 pp, medium confidence
182	Granite 4.2 3B	24.1%	measured
183	Gemma 3 27B	23.9%	estimated ± 3.2 pp, medium confidence
184	Celeris-1	22.2%	estimated ± 3.2 pp, medium confidence
185	Llama 4 Scout	21.9%	estimated ± 3.2 pp, medium confidence
186	GLM-5.2	19.2%	estimated ± 2.9 pp, low confidence
187	Laguna M.1	17.3%	estimated ± 2.9 pp, low confidence
188	GPT-4.1 mini	17.3%	estimated ± 3.8 pp, medium confidence
189	Laguna XS.2	16.9%	estimated ± 2.9 pp, low confidence
190	GLM-4.5	16.4%	estimated ± 2.9 pp, low confidence
191	LFM2.5-2.6B	15.2%	estimated ± 3.2 pp, medium confidence
192	Mellum2-12B-A2.5B-Instruct	11.4%	estimated ± 5.4 pp, low confidence
193	Grok Code Fast 1	10.5%	estimated ± 2.9 pp, low confidence
194	MiniCPM5-1B	10.0%	estimated ± 5.4 pp, low confidence
195	LLaDA2.2-mini	8.0%	estimated ± 5.4 pp, low confidence
196	Claude Haiku 4.5	1.0%	estimated ± 2.9 pp, low confidence

SciCode leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General