AA-SciCode leaderboard

A display-only Artificial Analysis SciCode score.

Models ranked by their published Artificial Analysis SciCode value

Rank	Model	Provider	Task success rate
1	Claude Fable 5.1	Anthropic	63.1
2	Claude Fable 5	Anthropic	61.0
3	Kimi K3	Moonshot AI	59.5
4	GLM-5.3	Z.AI	59.0
5	Muse Spark 1.1	Meta	58.8
5	Muse Spark 1.3	Meta	58.8
7	Gemini 3.1 Pro	Google	58.7
8	Muse Spark 1.2	Meta	57.4
9	Gemini 3.7 Flash	Google	57.2
10	GPT-5.6 Sol	OpenAI	57.1
11	Gemini 3.8 Flash	Google	56.6
12	GPT-6 Astra	OpenAI	56.5
12	Grok 4.6	xAI	56.5
14	Claude Opus 5	Anthropic	56.4
15	GPT-5.5	OpenAI	55.8
16	GPT-5.6 Terra	OpenAI	55.0
16	Grok 4.5	xAI	55.0
18	Claude Opus 4.8	Anthropic	54.4
19	Claude Sonnet 5	Anthropic	54.3
20	Gemini 3.5 Flash	Google	53.9
21	GPT-5.6 Luna	OpenAI	53.6
22	Gemini 3.6 Flash	Google	53.4
23	GPT-5.4 mini	OpenAI	52.1
23	Qwen3.8 Max Preview	Alibaba	52.1
25	DeepSeek V4.1 Flash	DeepSeek	51.9
26	GLM-5.3-Flash	Z.AI	51.6
27	Kimi K2.6	Moonshot AI	51.5
28	GLM-5.2	Z.AI	51.2
29	DeepSeek V4 Pro 0813	DeepSeek	51.0
30	MiMo-V2.5-Pro	Xiaomi	50.6
30	Qwen3.8-Flash-Next	Alibaba	50.6
32	DeepSeek V4 Flash 0731	DeepSeek	50.3
33	MiniMax M2.7	MiniMax	50.1
34	TMInkling-Small	Thinking Machines Lab	49.7
35	Qwen3.7 Max	Alibaba	49.5
36	Hy3	Tencent	48.6
36	Hy3 Preview	Tencent	48.6
38	Grok 4.3	xAI	48.3
39	MCQuasar 438B	Multiverse Computing	48.1
40	Kimi K2.7 Code	Moonshot AI	47.8
41	GPT-5.4 nano	OpenAI	47.2
42	MiniMax M3	MiniMax	47.1
43	TMInkling	Thinking Machines Lab	47.0
44	Qwen3.8-27B	Alibaba	46.6
45	Gemini 2.5 Pro	Google	46.3
46	Qwen3.7 Plus	Alibaba	46.1
47	APApodex 1.1	Apodex	45.5
47	APApodex 1.1 Mini	Apodex	45.5
47	Gemma 4 31B	Google	45.5
50	Muse Glimmer 30B	Meta	44.9
51	GLM-5.1	Z.AI	44.8
52	Solar Pro 4	Upstage	44.6
53	Ling 3.0 Flash VL	InclusionAI	44.2
54	Step 3.7 Flash	StepFun	43.9
55	Qwen3.6-27B	Alibaba	42.8
56	K-EXAONE 2.0	LG AI Research	42.0
56	Ling 3.0 Flash	InclusionAI	42.0
56	Ling 3.0 Flash FP8	InclusionAI	42.0
59	Gemini 3.5 Flash-Lite	Google	41.3
60	STA.X K2	SK Telecom	41.0
61	Trinity-Large-Preview	Arcee AI	40.6
61	Trinity-Large-Thinking	Arcee AI	40.6
63	Nemotron 3 Ultra	NVIDIA	40.3
64	Mistral Medium 3.5 128B	Mistral	40.2
65	Gemma 4 26B A4B	Google	40.0
66	Qwen3.5-122B-A10B	Alibaba	39.7
67	GPT-OSS 20B	OpenAI	38.9
68	Mistral Small 4	Mistral	38.8
68	Mistral Small 4 (Reasoning)	Mistral	38.8
68	North Mini Code	Cohere	38.8
71	Command A+	Cohere	38.5
72	Granite 4.2 30B	IBM	37.8
73	Mistral Large 3	Mistral	36.6
73	Qwen3.6-35B-A3B	Alibaba	36.6
75	Nemotron 3 Super 100B	NVIDIA	36.2
76	DeepSeek V3	DeepSeek	35.8
77	GPT-OSS 120B	OpenAI	34.0
78	Nemotron 3.5 Lightning 30B A3B NVFP4	NVIDIA	32.1
79	Llama 4 Maverick	Meta	31.7
80	Granite 4.2 8B	IBM	31.5
81	Nemotron 3 Nano 30B	NVIDIA	30.6
82	OPMiniCPM5-2B	OpenBMB	26.3
83	Solar Pro 3	Upstage	25.5
84	Granite 4.2 3B	IBM	25.3
85	Ling 3.0 Tiny	InclusionAI	24.2
86	Gemma 3 27B	Google	23.3
87	CECeleris-1	Celeris	21.6
88	Llama 4 Scout	Meta	21.3
89	LFM2.5-2.6B	LiquidAI	14.4

Evidence key: Observed

Rows are ordered by the value Artificial Analysis SciCode Benchmark Leaderboard published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.