SciCode — AI model leaderboard

AI models ranked by SciCode, an aggregated third-party benchmark from artificial_analysis. Higher is better. Cross-referenced against our first-party meo scores and Effective Value (𝕍).

Ranking 82 models across the full field · as of 2026-06-07.

#	Model	Lab	SciCode
1	Google: Gemini 3.1 Pro Preview	google	58.9%
2	OpenAI: GPT-5.4	openai	56.6%
3	OpenAI: GPT-5.5	openai	56.1%
4	OpenAI: GPT-5.2-Codex	openai	54.6%
5	Anthropic: Claude Opus 4.7	anthropic	54.5%
6	Anthropic: Claude Opus 4.8	anthropic	53.5%
7	OpenAI: GPT-5.3-Codex	openai	53.2%
8	Google: Gemini 3.5 Flash	google	53.1%
9	OpenAI: GPT-5.2 Chat	openai	52.1%
10	Xiaomi: MiMo-V2.5-Pro	xiaomi	50.2%
11	DeepSeek: DeepSeek V4 Pro	deepseek	50.0%
12	OpenAI: GPT-5.4 Mini	openai	49.9%
13	Google: Gemini 3 Flash Preview	google	49.9%
14	Qwen: Qwen3.7 Max	qwen	48.8%
15	xAI: Grok 4.3	x-ai	47.3%
16	MiniMax: MiniMax M2.7	minimax	47.0%
17	Anthropic: Claude Sonnet 4.6	anthropic	46.9%
18	OpenAI: GPT-5.4 Nano	openai	46.9%
19	OpenAI: o4 Mini	openai	46.5%
20	Qwen: Qwen3.7 Plus	qwen	45.5%
21	MiniMax: MiniMax M3	minimax	45.4%
22	DeepSeek: DeepSeek V4 Flash	deepseek	44.9%
23	Z.ai: GLM 5.1	z-ai	43.8%
24	Z.ai: GLM 5 Turbo	z-ai	43.6%
25	Z.ai: GLM 5V Turbo	z-ai	43.5%
26	Google: Gemma 4 31B	google	43.4%
27	OpenAI: GPT-5.1	openai	43.3%
28	OpenAI: GPT-5	openai	42.9%
29	Google: Gemini 2.5 Pro	google	42.8%
30	OpenAI: GPT-5.1-Codex-Mini	openai	42.6%
31	inclusionAI: Ring-2.6-1T	inclusionai	42.4%
32	Qwen: Qwen3.5 397B A17B	qwen	42.0%
33	Qwen: Qwen3.5-122B-A10B	qwen	42.0%
34	Google: Gemini 3.1 Flash Lite	google	41.9%
35	Tencent: Hy3 preview	tencent	41.2%
36	OpenAI: o3	openai	41.0%
37	OpenAI: GPT-5 Codex	openai	40.9%
38	Qwen: Qwen3.6 Plus	qwen	40.7%
39	OpenAI: GPT-4.1 Mini	openai	40.4%
40	OpenAI: GPT-5.1-Codex	openai	40.2%
41	StepFun: Step 3.7 Flash	stepfun	40.0%
42	Google: Gemma 4 26B A4B (free)	google	40.0%
43	OpenAI: o3 Mini	openai	39.9%
44	Qwen: Qwen3.6 27B	qwen	39.8%
45	OpenAI: o3 Mini High	openai	39.8%
46	Mistral: Mistral Medium 3.5	mistralai	39.6%
47	OpenAI: GPT-5 Mini	openai	39.2%
48	Prime Intellect: INTELLECT-3	prime-intellect	39.1%
49	OpenAI: gpt-oss-120b	openai	38.9%
50	Inception: Mercury 2	inception	38.7%
51	StepFun: Step 3.5 Flash	stepfun	38.5%
52	Kwaipilot: KAT-Coder-Pro V2	kwaipilot	38.3%
53	OpenAI: GPT-4.1	openai	38.1%
54	inclusionAI: Ling-2.6-1T	inclusionai	37.0%
55	OpenAI: GPT-5 Nano	openai	36.6%
56	Arcee AI: Trinity Large Thinking	arcee-ai	36.1%
57	Qwen: Qwen3.6 35B A3B	qwen	35.8%
58	OpenAI: o1	openai	35.8%
59	OpenAI: gpt-oss-20b	openai	34.4%
60	OpenAI: GPT-4o	openai	33.3%
61	OpenAI: GPT-4o (2024-08-06)	openai	33.1%
62	Meta: Llama 4 Maverick	meta-llama	33.1%
63	Qwen: Qwen3 Coder Next	qwen	32.3%
64	OpenAI: GPT-4 Turbo	openai	31.9%
65	OpenAI: GPT-4o (2024-05-13)	openai	30.9%
66	Google: Gemini 2.5 Flash	google	29.1%
67	Google: Gemini 2.5 Flash Lite Preview 09-2025	google	28.5%
68	Cohere: Command A	cohere	28.1%
69	Qwen: Qwen3.5-9B	qwen	27.5%
70	inclusionAI: Ling-2.6-flash	inclusionai	27.1%
71	Reka Flash 3	rekaai	26.7%
72	Microsoft: Phi 4	microsoft	26.0%
73	Xiaomi: MiMo-V2-Flash	xiaomi	25.9%
74	OpenAI: GPT-4.1 Nano	openai	25.9%
75	Upstage: Solar Pro 3	upstage	24.7%
76	OpenAI: GPT-4o-mini	openai	22.9%
77	IBM: Granite 4.1 8B	ibm-granite	21.8%
78	Google: Gemma 3 27B	google	21.2%
79	Google: Gemma 3 12B	google	17.4%
80	Meta: Llama 4 Scout	meta-llama	17.0%
81	Microsoft: Phi 4 Mini Instruct	microsoft	10.8%
82	Google: Gemma 3 4B	google	7.3%

Artificial Analysis (artificialanalysis.ai). Redistribution requires an AA commercial license.

← All rankings Methodology & 𝕍 →