AI Coding 模型排行榜

评分区间 0–100，来源于 Benchmark、编辑部实测与社区反馈的综合。

#	模型	Provider	综合	Coding	Agent	前端	上下文	性价比	趋势
1	GLM-5.3-Flash	智谱 AI	69	72	58	72	—	82	–
2	GPT-5.6 Luna (max)	OpenAI	66	71	47	71	—	72	–
3	DeepSeek V4 Flash 0731 (Reasoning, Max Effort)	DeepSeek	64	69	48	69	—	72	–
4	GPT-5.6 Luna (xhigh)	OpenAI	64	69	44	69	—	72	–
5	DeepSeek V4 Flash Vision (Reasoning, Max Effort)	DeepSeek	63	65	53	65	—	72	–
6	Gemini 3.7 Flash (high)	Google	68	76	45	76	—	60	–
7	GLM-5.3 (max)	智谱 AI	70	75	59	75	—	60	–
8	Muse Spark 1.2 (xhigh)	Meta	65	72	49	72	—	60	–
9	Gemini 3.7 Flash (medium)	Google	66	72	45	72	—	60	–
10	Muse Spark 1.1 (xhigh)	Meta	64	71	40	71	—	60	–
11	Gemini 3.7 Flash (low)	Google	64	71	42	71	—	60	–
12	Gemini 3.6 Flash (high)	Google	63	69	41	69	—	60	–
13	DeepSeek V4 Pro 0813 (Reasoning, Max Effort)	DeepSeek	64	69	50	69	—	60	–
14	GLM-5.2 (max)	智谱 AI	63	69	46	69	—	60	–
15	Qwen3.8 27B (xhigh)	Alibaba	63	68	51	68	—	60	–
16	Grok 4.6 (high)	SpaceXAI	70	77	59	77	—	48	–
17	Grok 4.6 (xhigh)	SpaceXAI	69	76	57	76	—	48	–
18	Grok 4.6 (medium)	SpaceXAI	68	74	56	74	—	48	–
19	Grok 4.5 (high)	SpaceXAI	65	72	49	72	—	48	–
20	Qwen3.8 2.4T A95B	Alibaba	66	72	57	72	—	48	–
21	Qwen3.8 Max	Alibaba	66	72	58	72	—	48	–
22	Gemini 3.5 Flash (high)	Google	62	70	40	70	—	48	–
23	Grok 4.6 (low)	SpaceXAI	61	66	48	66	—	48	–
24	Qwen3.7 Max	Alibaba	58	66	31	66	—	48	–
25	Claude Opus 5 (Adaptive Reasoning, Xhigh Effort)	Anthropic	69	77	58	77	—	35	–
26	GPT-5.6 Terra (max)	OpenAI	68	77	50	77	—	35	–
27	Claude Opus 5 (Adaptive Reasoning, High Effort)	Anthropic	69	77	56	77	—	35	–
28	Claude Fable 5 (Adaptive Reasoning, Max Effort, Opus 4.8 Fallback)	Anthropic	70	77	57	77	—	35	–
29	GPT-5.6 Sol (medium)	OpenAI	67	76	48	76	—	35	–
30	Kimi K3 (max)	Moonshot AI	68	76	54	76	—	35	–
31	GPT-5.5 (xhigh)	OpenAI	66	75	47	75	—	35	–
32	Claude Opus 4.8 (Adaptive Reasoning, Max Effort)	Anthropic	65	74	49	74	—	35	–
33	Claude Opus 5 (Adaptive Reasoning, Medium Effort)	Anthropic	65	74	50	74	—	35	–
34	Claude Opus 4.7 (Adaptive Reasoning, Max Effort)	Anthropic	64	74	46	74	—	35	–
35	Kimi K3 (low)	Moonshot AI	62	72	40	72	—	35	–
36	GPT-5.5 (high)	OpenAI	64	72	46	72	—	35	–
37	Claude Sonnet 5 (Adaptive Reasoning, Max Effort)	Anthropic	65	72	50	72	—	35	–
38	GPT-5.5 (medium)	OpenAI	62	72	39	72	—	35	–
39	GPT-5.4 (xhigh)	OpenAI	63	71	44	71	—	35	–
40	GPT-5.6 Terra (xhigh)	OpenAI	64	71	47	71	—	35	–
41	GPT-5.6 Sol (low)	OpenAI	62	70	42	70	—	35	–
42	Gemini 3.1 Pro Preview	Google	56	69	23	69	—	35	–
43	GPT-5.6 Terra (high)	OpenAI	60	67	44	67	—	35	–
44	Claude Opus 5 (Adaptive Reasoning, Low Effort)	Anthropic	59	67	42	67	—	35	–
45	Claude Sonnet 5 (Non-reasoning, High Effort)	Anthropic	57	66	34	66	—	35	–
46	GPT-5.6 Sol (Non-reasoning)	OpenAI	57	65	36	65	—	35	–
47	GPT-5.6 Sol (xhigh)	OpenAI	69	78	54	78	—	35	–
48	Claude Opus 5 (Adaptive Reasoning, Max Effort)	Anthropic	70	78	59	78	—	35	–
49	GPT-5.6 Sol (max)	OpenAI	70	77	58	77	—	35	–
50	GPT-5.6 Sol (high)	OpenAI	68	77	51	77	—	35	–

GLM-5.3-Flash

智谱 AI · 上下文 —

说明：评分为编辑团队基于公开基准（Benchmark）、编辑部实测与社区反馈的综合估算，非官方数据；点击「按分类查看」可切换排序维度。