Where do these scores come from?

Every score comes from a public benchmark or official machine-readable file, with a source link and date. This site does not invent scores, and missing values are not treated as zero.

How is the overall score calculated?

Each source is min-max normalized within itself, then weighted: Artificial Analysis 28%, LMArena Text Arena 28%, LiveBench 22%, coding composite 15%, reasoning composite 7%. Missing sources are dropped and the rest are renormalized. A model needs at least two headline sources (or three partials) to enter the overall board. When several variants of one family exist, only the best overall variant is shown.

Are site prices part of the benchmarks?

No. Benchmark scores come from third-party public boards. CNY prices are this site’s catalogue reference rates (CNY 1 = 1,000 points) and are settled on actual usage. They are shown separately so billing is not confused with quality.

How often is this updated?

A collector runs automatically every day. The upstream boards may not publish a new snapshot daily, so each page shows both this site’s collection time and the source publication date.

88Z.ai / RANKINGS

Open-source model leaderboard

Models marked open-source or non-proprietary in the source data, ranked with the same composite method. License labels follow the original sources.

Last collected by this site: 2026-11-14 06:41 (UTC+8, Asia/Shanghai). Source publication dates are in the table or the citations below.

Models marked open-source or non-proprietary in the source data, ranked with the same composite method. License labels follow the original sources.

177 model families (best published variant per family)

#	Model	Overall	AA	LMArena	LiveBench	Coding	Reasoning	Our price / 1M tokens
1	DeepSeek v4.1 Flash 20260910 (Max)DeepSeek	85.6	—	1,475	—	71.7	93.6	—
2	GLM 5.2 (Max)zai	84.9	—	1,475	—	70.0	92.3	in ¥0.06 / out ¥58.48
3	GLM 5Z-AI	82.5	—	1,457	—	70.8	85.5	in ¥8.6 / out ¥27.52
4	DeepSeek V4 Pro High 20260813DeepSeek	82.0	—	1,464	—	66.5	86.5	—
5	DeepSeek R1 0528DeepSeek	80.7	—	1,422	—	80.3	77.7	in ¥4.3 / out ¥18.49
6	GLM 5.1zai	79.8	—	1,465	—	57.0	90.4	in ¥12.04 / out ¥37.84
7	Kimi k2.6Moonshot	79.5	—	1,461	—	57.2	91.1	in ¥5.59 / out ¥29.33
8	Mimo v2.5 Proxiaomi	78.9	—	1,466	—	53.3	91.1	in ¥3.74 / out ¥7.48
9	Hy3Tencent	78.8	—	1,454	—	56.7	91.9	in ¥1.14 / out ¥4.54
10	DeepSeek v4.1 Flash 1017DeepSeek	77.6	—	—	70.3	85.0	78.8	in ¥0.33 / out ¥4.3
11	DeepSeek V4 Pro Preview (High)DeepSeek	76.8	—	1,455	—	51.2	88.9	—
12	DeepSeek v3.2DeepSeek	76.2	—	1,425	—	61.6	82.8	in ¥2.24 / out ¥3.27
13	DeepSeek V4 ProDeepSeek	76.1	—	1,458	—	49.3	85.5	in ¥7.74 / out ¥12.9
14	GLM 4.5Z.ai	76.0	—	1,411	—	66.7	80.6	in ¥5.16 / out ¥18.92
15	Kimi k2.5 (Thinking)Moonshot	75.4	—	1,450	—	47.7	89.6	in ¥4.3 / out ¥21.5
16	MiniMax m2.5MiniMax	74.6	—	1,391	—	69.8	77.7	in ¥2.32 / out ¥8.17
17	Beam 1.0 501b A23breflection	73.9	—	1,434	—	47.7	89.9	in ¥2.58 / out ¥8.6
18	DeepSeek R1DeepSeek	73.7	—	1,398	—	63.2	80.7	—
19	GLM 4.7zai	73.6	—	1,441	—	47.5	83.0	in ¥5.16 / out ¥18.92
20	DeepSeek V4 Flash Preview (High)DeepSeek	73.5	—	1,439	—	47.0	85.2	—
21	Mimo v2.5xiaomi	73.2	—	1,434	—	47.9	84.6	in ¥3.44 / out ¥17.2
22	GLM 4.6Z.ai	73.1	—	1,424	—	52.0	81.6	in ¥4.3 / out ¥17.2
23	qwen3.5 397b A17bAlibaba	72.7	—	1,441	—	43.0	86.6	in ¥5.16 / out ¥30.96
24	GLM 5.3 (Max)Zhipu	72.5	32	1,482	—	71.9	94.6	in ¥0.72 / out ¥30.96
25	DeepSeek V3 0324DeepSeek	71.8	—	1,395	—	61.0	73.7	in ¥2.49 / out ¥9.8
26	Kimi k2.5 InstantMoonshot	71.7	—	1,431	—	43.7	85.2	—
27	Gemma 4 26b A4bGoogle	71.1	—	1,437	—	37.6	89.6	in ¥0.58 / out ¥1.94
28	MiniMax m2.7MiniMax	69.4	—	1,415	—	42.7	82.4	in ¥1.81 / out ¥7.22
29	Nvidia Nemotron 3.5 Super Vl 120b A12b Nvfp4NVIDIA	69.4	—	1,414	—	43.1	81.4	—
30	Muse GlimmerMeta	69.1	—	1,425	—	37.0	85.4	—
31	Kimi k3.1 ProMoonshot	68.8	46	—	68.7	67.2	67.5	in ¥17.2 / out ¥86
32	Kimi K2 Thinking TurboMoonshot	68.2	—	1,430	—	32.7	84.4	—
33	qwen3.5 122b A10bAlibaba	68.2	—	1,416	—	37.8	82.7	in ¥2.24 / out ¥17.89
34	Hy3 PreviewTencent	67.5	—	1,409	—	37.5	83.4	in ¥1.55 / out ¥5.16
35	qwen3.5 27bAlibaba	67.3	—	1,408	—	37.4	82.4	in ¥2.24 / out ¥22.36
36	Qwen3 Coder 480b A35b InstructAlibaba	67.2	—	1,387	—	47.5	75.1	—
37	qwen3.8 27bAlibaba	65.8	—	1,434	54.1	68.5	65.2	in ¥1.29 / out ¥16.13
38	Mimo v2.6 Pro小米集团	65.5	30	1,480	66.1	67.4	44.7	in ¥3.74 / out ¥7.48
39	MiniMax m2.1 PreviewMiniMax	65.2	—	1,384	—	40.9	77.2	—
40	Mimo v2.6 Flash小米集团	64.8	27	1,448	—	74.0	59.2	in ¥1.2 / out ¥2.41
41	MiniMax M2MiniMax	64.2	—	1,344	—	53.3	71.8	—
42	Mimo V2 Flash Non Thinking (Thinking)xiaomi	63.6	—	1,392	—	34.0	75.0	—
43	GLM 5.3 FlashZhipu	63.3	28	1,475	60.7	70.9	53.3	in ¥0.34 / out ¥0.67
44	Mistral Large 3Mistral	62.5	—	1,415	—	21.1	79.1	—
45	Qwen3 32bAlibaba	62.4	—	1,347	—	43.1	78.6	in ¥0.69 / out ¥2.41
46	Mimo V2 Flash (Thinking)xiaomi	61.4	—	1,386	—	28.9	74.6	—
47	qwen3.5 35b A3bAlibaba	61.2	—	1,394	—	23.8	78.3	in ¥1.2 / out ¥8.6
48	Step 5 Preview阶跃星辰	59.8	29	1,449	54.9	66.9	59.2	in ¥8.6 / out ¥23.22
49	Trinity Large ThinkingArcee AI	57.6	—	1,367	—	21.6	76.2	—
50	GPT Oss 120bOpenAI	57.1	—	1,352	—	25.1	75.9	in ¥1.29 / out ¥5.16
51	MiniMax M3稀宇科技	56.2	19	1,436	—	53.9	66.3	in ¥2.58 / out ¥10.32
52	Inklingthinky	55.2	—	1,442	44.3	42.5	64.2	in ¥8.6 / out ¥34.83
53	MiniMax m3.1 Flash稀宇科技	54.5	29	—	63.7	67.8	57.2	in ¥1.29 / out ¥5.16
54	Mistral Large 4Mistral	54.0	—	1,420	40.9	49.2	73.1	in ¥6.02 / out ¥18.06
55	Qwq 32bAlibaba	53.7	—	1,336	—	20.5	73.1	—
56	Inkling SmallThinking Machines	50.8	13	1,405	—	44.0	86.9	in ¥3.87 / out ¥10.32
57	Granite 4.1 8bIBM	48.3	—	1,304	—	16.0	66.1	—
58	Mistral Medium 3.5Mistral	43.6	5	1,426	—	25.1	82.9	in ¥12.9 / out ¥64.5
59	qwen2.5 Coder 32b InstructAlibaba	41.5	—	1,271	—	7.6	58.6	in ¥5.68 / out ¥8.6
60	Granite 4.2 8bIBM	31.7	5	1,289	—	—	—	in ¥0.52 / out ¥2.15
61	Granite 4.2 3bIBM	31.7	5	1,289	—	—	—	—
62	Gemma 4 31bGoogle	31.3	4	1,452	32.4	19.5	44.6	in ¥1.72 / out ¥3.44
63	DeepSeek V4 FlashDeepSeek	—	—	1,436	—	—	82.7	in ¥0.77 / out ¥1.55
64	Nvidia Nemotron 3 Ultra 550b A55b Nvfp4NVIDIA	—	—	1,426	—	—	85.4	—
65	DeepSeek v3.2 Exp (Thinking)DeepSeek	—	—	1,425	—	—	82.1	in ¥2.32 / out ¥3.53
66	Qwen3 235b A22b Instruct 2507Alibaba	—	—	1,422	—	—	81.4	in ¥1.2 / out ¥6.88
67	Kimi K2 Preview 0905Moonshot	—	—	1,418	—	—	81.3	—
68	Kimi K2 Preview 0711Moonshot	—	—	1,418	—	—	77.0	—
69	DeepSeek v3.1DeepSeek	—	—	1,417	—	—	80.8	in ¥2.15 / out ¥8.17
70	DeepSeek v3.1 Terminus (Thinking)DeepSeek	—	—	1,416	—	—	—	in ¥2.32 / out ¥8.6
71	Qwen3 Vl 235b A22b InstructAlibaba	—	—	1,413	—	—	80.3	in ¥1.81 / out ¥16.34
72	Qwen3 235b A22b Non Thinking (Thinking)Alibaba	—	—	1,402	—	—	77.4	—
73	Longcat Flash Chatmeituan	—	—	1,401	—	—	80.9	—
74	Qwen3 235b A22b Thinking 2507Alibaba	—	—	1,400	—	—	78.4	in ¥3.87 / out ¥30.1
75	Qwen3 Next 80b A3b InstructAlibaba	—	—	1,399	—	—	81.0	in ¥0.77 / out ¥9.46
76	Qwen3 Vl 235b A22b ThinkingAlibaba	—	—	1,395	—	—	78.6	—
77	Step 3.5 FlashStepFun	—	—	1,393	—	—	79.7	in ¥0.86 / out ¥2.58
78	Qwen3 30b A3b Instruct 2507Alibaba	—	—	1,382	—	—	75.4	in ¥0.86 / out ¥2.58
79	Trinity Large PreviewOther	—	—	1,379	—	—	72.9	—
80	GLM 4.6vzai	—	—	1,378	—	—	—	in ¥2.58 / out ¥7.74
81	Qwen3 235b A22bAlibaba	—	—	1,375	—	—	77.5	—
82	GLM 4.5 Airzai	—	—	1,373	—	—	76.7	in ¥1.72 / out ¥9.46
83	Qwen3 Next 80b A3b ThinkingAlibaba	—	—	1,369	—	—	77.5	—
84	GLM 4.7 Flashzai	—	—	1,365	—	—	72.9	in ¥0.52 / out ¥3.44
85	MiniMax M1MiniMax	—	—	1,364	—	—	74.2	—
86	Nvidia Nemotron 3 Super 120b A12bNVIDIA	—	—	1,360	—	—	74.2	—
87	Mistral Small 3.2 2506Mistral	—	—	1,357	—	—	69.2	—
88	Intellect 3Prime Intellect	—	—	1,356	—	—	74.8	—
89	Command A 03 2025Cohere	—	—	1,354	—	—	65.0	—
90	GLM 4.5vzai	—	—	1,352	—	—	72.7	in ¥5.16 / out ¥15.48
91	Nvidia Nemotron 3.5 Lightning 30b A3b Nvfp4NVIDIA	—	—	1,351	—	—	75.1	—
92	Step 3StepFun	—	—	1,349	—	—	72.7	—
93	Llama 3.1 Nemotron Ultra 253b V1NVIDIA	—	—	1,348	—	—	75.9	—
94	Llama 3.3 Nemotron Super 49b v1.5NVIDIA	—	—	1,343	—	—	77.2	—
95	Ling Flash 2.0ant-group	—	—	1,343	—	—	70.3	—
96	Granite 4.2 30bIBM	—	—	1,340	—	—	—	—
97	Llama 3.1 405b Instruct Bf16Meta	—	—	1,336	—	—	65.8	—
98	Llama 3.1 405b Instruct Fp8Meta	—	—	1,334	—	—	66.4	—
99	Olmo 3.1 32b InstructAllen Institute for AI	—	—	1,329	—	—	63.1	—
100	Qwen3 30b A3bAlibaba	—	—	1,327	—	—	71.3	in ¥1.03 / out ¥4.3
101	Ring Flash 2.0InclusionAI	—	—	1,322	—	—	69.1	—
102	Molmo 2 8ballenai	—	—	1,321	—	—	—	—
103	GPT Oss 20bOpenAI	—	—	1,318	—	—	68.8	in ¥0.26 / out ¥1.12
104	Nvidia Nemotron 3 Nano 30b A3b Bf16NVIDIA	—	—	1,314	—	—	70.6	—
105	Athene 70b 0725Other	—	—	1,307	—	—	57.6	—
106	Olmo 3 32b Thinkallenai	—	—	1,306	—	—	65.0	—
107	Mistral Small 3.1 24b Instruct 2503Mistral	—	—	1,303	—	—	59.7	—
108	Llama 3.1 70b InstructMeta	—	—	1,293	—	—	58.6	in ¥6.19 / out ¥6.19
109	Jamba 1.5 LargeOther	—	—	1,290	—	—	54.8	—
110	Granite 4.0 H SmallIBM	—	—	1,288	—	—	60.6	—
111	Olmo 3.1 32b ThinkAllen Institute for AI	—	—	1,286	—	—	63.7	—
112	Gemma 2 9b It SimpoGoogle	—	—	1,281	—	—	49.1	—
113	Nemotron 4 340b InstructNVIDIA	—	—	1,277	—	—	56.0	—
114	Llama 3 70b InstructMeta	—	—	1,277	—	—	57.0	—
115	Command R+ 08 2024Cohere	—	—	1,276	—	—	52.8	in ¥21.5 / out ¥86
116	Mistral Small 24b Instruct 2501Mistral	—	—	1,274	—	—	57.4	in ¥0.43 / out ¥0.69
117	Aya Expanse 32bCohere	—	—	1,267	—	—	53.1	—
118	Command R+Cohere	—	—	1,262	—	—	50.1	—
119	Phi 4Microsoft	—	—	1,256	—	—	57.9	in ¥0.6 / out ¥1.2
120	Olmo 2 32b Instruct 0325allenai	—	—	1,252	—	—	52.0	—
121	Command R 08 2024Cohere	—	—	1,251	—	—	49.1	in ¥1.29 / out ¥5.16
122	Jamba 1.5 (Mini)Other	—	—	1,240	—	—	45.8	—
123	Mixtral 8x22b Instruct v0.1Mistral	—	—	1,230	—	—	52.3	—
124	Command RCohere	—	—	1,227	—	—	44.4	—
125	Llama 3 8b InstructMeta	—	—	1,224	—	—	46.8	—
126	Aya Expanse 8bCohere	—	—	1,223	—	—	48.1	—
127	Zephyr Orpo 141b A35b v0.1Other	—	—	1,213	—	—	47.4	—
128	Yi 1.5 34b ChatOther	—	—	1,213	—	—	49.8	—
129	Llama 3.1 8b InstructMeta	—	—	1,211	—	—	46.1	in ¥0.17 / out ¥0.34
130	Granite 3.1 8b InstructIBM	—	—	1,209	—	—	46.4	—
131	Phi 3 Medium 4k InstructMicrosoft	—	—	1,198	—	—	50.4	—
132	Mixtral 8x7b Instruct v0.1Mistral	—	—	1,197	—	—	46.5	—
133	Wizardlm 70bMicrosoft	—	—	1,185	—	—	41.3	—
134	Granite 3.0 8b InstructIBM	—	—	1,183	—	—	47.7	—
135	Openchat 3.5Other	—	—	1,183	—	—	36.4	—
136	Openchat 3.5 0106Other	—	—	1,183	—	—	41.6	—
137	Snowflake Arctic InstructOther	—	—	1,180	—	—	42.3	—
138	Granite 3.1 2b InstructIBM	—	—	1,179	—	—	47.6	—
139	Openhermes 2.5 Mistral 7bOther	—	—	1,176	—	—	40.6	—
140	Phi 3 Small 8k InstructMicrosoft	—	—	1,171	—	—	46.9	—
141	Starling Lm 7b BetaOther	—	—	1,171	—	—	41.3	—
142	Llama 2 70b ChatMeta	—	—	1,171	—	—	38.1	—
143	Starling Lm 7b AlphaOther	—	—	1,168	—	—	36.6	—
144	Nous Hermes 2 Mixtral 8x7b DpoOther	—	—	1,165	—	—	31.3	—
145	Granite 3.0 2b InstructIBM	—	—	1,157	—	—	43.2	—
146	Llama 2 70b Steerlm ChatNVIDIA	—	—	1,155	—	—	34.6	—
147	Qwq 32b PreviewAlibaba	—	—	1,154	—	—	49.2	—
148	Solar 10.7b Instruct v1.0upstage	—	—	1,153	—	—	33.8	—
149	Dolphin 2.2.1 Mistral 7bOther	—	—	1,152	—	—	36.4	—
150	Mpt 30b ChatOther	—	—	1,151	—	—	31.8	—
151	Wizardlm 13bMicrosoft	—	—	1,150	—	—	27.0	—
152	Mistral 7b Instruct v0.2Mistral	—	—	1,150	—	—	36.6	—
153	Phi 3 Mini 4k Instruct June 2024Microsoft	—	—	1,143	—	—	46.9	—
154	Vicuna 13bOther	—	—	1,142	—	—	29.9	—
155	Llama 2 13b ChatMeta	—	—	1,142	—	—	34.0	—
156	Code Llama 34b InstructMeta	—	—	1,137	—	—	34.0	—
157	Zephyr 7b BetaOther	—	—	1,131	—	—	29.7	—
158	Phi 3 Mini 128k InstructMicrosoft	—	—	1,130	—	—	38.6	—
159	Phi 3 Mini 4k InstructMicrosoft	—	—	1,128	—	—	40.1	—
160	Zephyr 7b AlphaOther	—	—	1,127	—	—	—	—
161	Stripedhyena Nous 7bOther	—	—	1,122	—	—	30.2	—
162	Code Llama 70b InstructMeta	—	—	1,119	—	—	—	—
163	Vicuna 7bOther	—	—	1,115	—	—	24.4	—
164	Smollm2 1.7b InstructOther	—	—	1,115	—	—	40.6	—
165	Mistral 7b InstructMistral	—	—	1,111	—	—	29.8	—
166	Llama 2 7b ChatMeta	—	—	1,108	—	—	30.3	—
167	Olmo 7b Instructallenai	—	—	1,074	—	—	25.2	—
168	Mpt 7b ChatOther	—	—	1,064	—	—	15.0	—
169	Chatglm3 6bZhipu	—	—	1,057	—	—	23.5	—
170	Rwkv 4 Raven 14bOther	—	—	1,043	—	—	14.6	—
171	Chatglm2 6bZhipu	—	—	1,025	—	—	12.7	—
172	Openassistant Pythia 12bOther	—	—	1,024	—	—	11.1	—
173	Fastchat T5 3bOther	—	—	993	—	—	4.7	—
174	Dolly 2.0 12bOther	—	—	983	—	—	9.9	—
175	Stablelm Tuned Alpha 7bOther	—	—	953	—	—	0	—
176	qwen3.8 Flash NextAlibaba	—	29	—	—	72.1	—	—
177	Nemotron 3 UltraNVIDIA	—	12	—	—	—	—	in ¥5.16 / out ¥20.64

Sources

LMArena Text Arena (style-controlled) — Official Arena leaderboard dataset, text_style_control / latest. Scores are Arena ratings (Bradley-Terry).
LMArena WebDev Arena — Official webdev / latest snapshot, measuring front-end and web-app battles.
LMArena Vision Arena (style-controlled) — Official vision_style_control / latest snapshot.
LMArena Text-to-Image Arena — Official text_to_image / latest snapshot.
LMArena Text-to-Video Arena — Official text_to_video / latest snapshot.
LMArena Agent Arena — Official agent / latest aggregate scores.
Artificial Analysis Intelligence Index — Intelligence Index, cost per task, time per task and first-token latency from the official public leaderboard page.
LiveBench 2.0 — Official CSV results. Overall uses the published formula: Long Agentic ×2, Design ×1.5, Coding ×1.5, other suites ×1, then a weighted mean over suites that have a score.
Aider Polyglot — Official polyglot_leaderboard.yml. The metric is pass_rate_2 on 225 cross-language exercises. Repeated submissions keep the best score.
SWE-bench Verified (mini-SWE-agent) — Official leaderboards.json, Verified board, resolved % under the mini-SWE-agent protocol so models are compared in a similar scaffold. Scores from other agents are not mixed into the composite coding score.
SuperCLUE Chinese general board — Total and section scores from the official monthly general-board spreadsheet.

FAQ

Where do these scores come from?: Every score comes from a public benchmark or official machine-readable file, with a source link and date. This site does not invent scores, and missing values are not treated as zero.
How is the overall score calculated?: Each source is min-max normalized within itself, then weighted: Artificial Analysis 28%, LMArena Text Arena 28%, LiveBench 22%, coding composite 15%, reasoning composite 7%. Missing sources are dropped and the rest are renormalized. A model needs at least two headline sources (or three partials) to enter the overall board. When several variants of one family exist, only the best overall variant is shown.
Are site prices part of the benchmarks?: No. Benchmark scores come from third-party public boards. CNY prices are this site’s catalogue reference rates (CNY 1 = 1,000 points) and are settled on actual usage. They are shown separately so billing is not confused with quality.
How often is this updated?: A collector runs automatically every day. The upstream boards may not publish a new snapshot daily, so each page shows both this site’s collection time and the source publication date.