💻 Coding

AI model rankings by coding capability — LiveCodeBench, SciCode, Terminal-Bench.

オープンソース100/100 件のモデル

#			アクセス
1	Claude Fable 5.1 Anthropic	99.3	試す
2	GPT-5.6 Sol OpenAI	94.8	試す
3	Claude Opus 5 Anthropic	94.8	試す
4	GPT-6 Astra OpenAI	94.6	試す
5	Muse Spark 1.3 Meta	94.0	試す
6	Grok 4.6 SpaceXAI	93.9	–
7	Gemini 3.8 Flash Google	93.8	試す
8	Qwen3.8 Max Alibaba	93.7	試す
9	Gemini 3.7 Flash Google	93.2	試す
10	GPT-5.6 Terra OpenAI	93.1	試す
11	Kimi K3 Kimi	92.7	試す
12	Claude Fable 5 Anthropic	92.3	試す
13	GLM-5.3 Z AI	91.9	試す
14	Qwen3.8-Flash-Next Alibaba	90.3	試す
15	Claude Opus 4.8 Anthropic	89.7	試す
16	GPT-5.5 OpenAI	89.7	試す
17	Muse Spark 1.2 Meta	88.9	試す
18	Qwen3.8 2.4T A95B Alibaba	88.7	試す
19	GLM 5.3 Flash Z AI	88.6	試す
20	Grok 4.5 SpaceXAI	88.6	試す
21	Claude Opus 4.7 Anthropic	88.2	試す
22	GPT-5.6 Luna OpenAI	87.5	試す
23	Claude Sonnet 5 Anthropic	87.5	試す
24	Muse Spark 1.1 Meta	87.4	試す
25	DeepSeek V4 Flash 0731 DeepSeek	85.6	試す
26	DeepSeek V4 Pro 0813 DeepSeek	85.5	試す
27	Gemini 3.6 Flash Google	85.5	試す
28	Gemini 3.5 Flash Google	85.1	試す
29	Qwen3.8 27B Alibaba	84.8	試す
30	GPT-5.4 OpenAI	84.7	試す
31	GLM-5.2 Z AI	84.3	試す
32	Gemini 3.1 Pro Preview Google	82.0	試す
33	DeepSeek V4 Flash Vision DeepSeek	81.7	試す
34	Qwen3.7 Max Alibaba	80.8	試す
35	Motif 3 Motif Technologies	79.9	–
36	K2 Horizon 375B A23B Institute of Foundation Models	78.2	–
37	Quasar 438B Multiverse Computing	77.5	–
38	Apodex 1.1 Apodex	77.4	–
39	Claude Sonnet 4.6 Anthropic	75.9	試す
40	Kimi K2.7 Code Kimi	75.8	試す
41	Kimi K2.6 Kimi	75.8	試す
42	Hy3 Tencent	74.2	–
43	MiMo-V2.5-Pro Xiaomi	74.2	–
44	Nex-N2-Pro Nex AGI	73.9	–
45	Ling-3.0-flash-VL InclusionAI	73.7	–
46	DeepSeek V4 Pro 0424 DeepSeek	73.4	試す
47	MiniMax-M3 MiniMax	73.3	試す
48	KAT Coder Pro V2 KwaiKAT	72.9	–
49	Muse Spark Meta	72.2	試す
50	Ling-3.0-flash-Fin InclusionAI	72.2	–
51	MiMo-V2.5 Xiaomi	70.6	–
52	Qwen3.7 Plus Alibaba	70.5	試す
53	DeepSeek V4 Flash 0420 DeepSeek	70.0	試す
54	GLM-5.1 Z AI	69.3	試す
55	GPT-5.4 nano OpenAI	69.2	試す
56	GPT-5.4 mini OpenAI	69.2	試す
57	Inkling Small Thinking Machines	68.5	–
58	Solar Pro 4 Upstage	68.4	–
59	Qwen3.6 Plus Alibaba	67.8	試す
60	JT-4.1 Flash 236B A21B China Mobile	67.6	–
61	Qwen3.6 27B Alibaba	67.4	試す
62	Inkling Thinking Machines	67.4	–
63	Ling 3.0 Flash InclusionAI	66.2	–
64	Grok Build 0.1 0616 SpaceXAI	66.2	–
65	MiniMax-M2.7 MiniMax	65.5	試す
66	Muse Glimmer Meta	64.6	試す
67	Gemini 3.5 Flash-Lite Google	64.5	試す
68	Nemotron 3 Ultra 550B A55B NVIDIA	63.6	試す
69	Claude 4.5 Sonnet Anthropic	61.6	試す
70	MiMo-V2-Flash Xiaomi	60.7	試す
71	Solar Open2 250B Upstage	60.5	–
72	Mistral Medium 3.5 Mistral	60.4	試す
73	Qwen3.5 397B A17B Alibaba	60.3	試す
74	LongCat 2.0 LongCat	59.9	–
75	GPT-5.1 OpenAI	59.7	試す
76	Kimi K2.5 Kimi	58.4	試す
77	Qwen3.5 122B A10B Alibaba	57.8	試す
78	Ring-2.6-1T InclusionAI	56.2	–
79	Gemma 4 31B Google	56.1	試す
80	GLM-4.7 Z AI	56.1	試す
81	K-EXAONE 2.0 0803 LG AI Research	55.8	–
82	Grok 4.3 SpaceXAI	55.4	–
83	GLM-4.6 Z AI	55.0	試す
84	Qwen3.6 35B A3B Alibaba	54.9	試す
85	DeepSeek V3.2 DeepSeek	54.5	試す
86	K2 Horizon 7B Institute of Foundation Models	54.2	–
87	A.X-K2 SK Telecom	54.0	–
88	GPT-5.5 Instant OpenAI	53.6	試す
89	Gemini 2.5 Pro Preview Google	53.6	試す
90	Step 3.7 Flash StepFun	53.3	試す
91	Claude 4.5 Haiku Anthropic	53.3	試す
92	DeepSeek V3.1 Terminus DeepSeek	52.4	試す
93	Gemma 4 26B A4B Google	51.8	試す
94	North Mini Code Cohere	50.3	試す
95	Nemotron 3 Super 120B A12B NVIDIA	49.7	試す
96	Qwen3.5 35B A3B Alibaba	48.7	試す
97	G9v3-39A5B AI9Stars	48.1	–
98	Qwen3 Coder Next Alibaba	47.4	試す
99	Gemini 3.1 Flash-Lite Google	46.4	試す
100	o1 OpenAI	46.2	試す

スコア算出方法 — AIM Index

● 能力 85%● 最新性 15%

SWE-bench VerifiedAider PolyglotLiveCodeBench

データソース (1)

Artificial Analysis ↗