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

AI coding model leaderboard

Covers coding battles, software-engineering repair, polyglot editing and the LiveBench coding suite.

Last collected by this site: 2026-11-13 06:41 (UTC+8, Asia/Shanghai). Source publication dates are in the table or the citations below.

Covers coding battles, software-engineering repair, polyglot editing and the LiveBench coding suite.

234 model families (best published variant per family)

#	Model	Coding score	WebDev	LiveBench	Aider	SWE-bench
1	Gemini 4 CarbonGoogle	100	—	91	—	—
2	GPT 5 (High)OpenAI	100	—	—	88	—
3	Claude Fable 5.2 (xHigh)Anthropic	100	1,836	—	—	—
4	Claude 4.5 OpusAnthropic	100	—	—	—	76.8
5	Claude 4.6 OpusAnthropic	98.2	—	—	—	75.6
6	Claude Fable 5.1Anthropic	98.2	—	89.7	—	—
7	GPT 6.1 TerraOpenAI	97.5	—	89.2	—	—
8	Claude Opus 5.5 (Max)Anthropic	97.2	1,814	—	—	—
9	O3 Pro (High)OpenAI	96.3	—	—	84.9	—
10	Gemini 3 Pro PreviewGoogle	96.2	—	—	—	74.2
11	GPT 6 AstraOpenAI	95.0	—	87.5	—	—
12	Gemini 2.5 Pro Preview 06 05 32k Think (Thinking)Google	94.2	—	—	83.1	—
13	Gemini 4 ArgonGoogle	92.5	—	85.7	—	—
14	O3 (High)OpenAI	92.1	—	—	81.3	—
15	Claude 4.5 SonnetAnthropic	92.0	—	—	—	71.4
16	Claude Sonnet 5.5 (xHigh)Anthropic	91.7	1,773	—	—	—
17	Kimi k2.5Moonshot	91.2	—	—	—	70.8
18	Muse Spark 1.3Meta	90.7	—	84.4	—	—
19	GPT 5.6 TerraOpenAI	90.4	—	84.2	—	—
20	Grok 4 (High)xAI	90.0	—	—	79.6	—
21	Gemini 2.5 Pro Preview 06 05 Default Think (Thinking)Google	89.5	—	—	79.1	—
22	O3 + GPT 4.1 (High)OpenAI	88.4	—	—	78.2	—
23	GPT 6.1 Sol (Max)OpenAI	87.9	1,743	—	—	—
24	Gemini 2.5 Pro Preview 05 06Google	86.8	—	—	76.9	—
25	Claude 4 OpusAnthropic	86.4	—	—	—	67.6
26	Muse Wonder Full (Max)Meta	85.3	1,723	—	—	—
27	DeepSeek v4.1 Flash 1017DeepSeek	85.0	—	80.4	—	—
28	Claude 4.5 HaikuAnthropic	85.0	—	—	—	66.6
29	DeepSeek v3.2 Exp ReasonerDeepSeek	83.6	—	—	74.2	—
30	Mistral Large 4 0Mistral	83.6	—	79.4	—	—
31	Claude 4 SonnetAnthropic	82.5	—	—	—	64.9
32	Gemini 4 HydrogenGoogle	82.3	—	78.5	—	—
33	Gemini 2.5 Pro Preview 03 25Google	82.1	—	—	72.9	—
34	Claude Haiku 5.5Anthropic	81.0	—	77.6	—	—
35	Claude Opus 5 (Max)Anthropic	81.0	1,691	—	—	—
36	O4 Mini (High)OpenAI	81.0	—	—	72	—
37	Claude Opus 4 20250514 32k Thinking (Thinking)Anthropic	81.0	—	—	72	—
38	Kimi k3.1 Pro (Max)Moonshot	80.6	1,688	—	—	—
39	DeepSeek R1 0528DeepSeek	80.3	—	—	71.4	—
40	Kimi K2 ThinkingMoonshot	80.2	—	—	—	63.4
41	Grok 4.6xAI	79.8	—	76.7	—	—
42	Claude Opus 4 20250514 No Think (Thinking)Anthropic	79.5	—	—	70.7	—
43	qwen3.8 Max 0902阿里巴巴	79.1	1,676	—	—	—
44	DeepSeek v3.2 Exp ChatDeepSeek	78.9	—	—	70.2	—
45	GPT 6 Sol (Max)OpenAI	78.7	1,673	—	—	—
46	DeepSeek v4.1 Flash 20261017 (Max)DeepSeek	77.5	1,664	—	—	—
47	Muse Wonder 1.0Meta	77.2	—	74.9	—	—
48	Kimi K3 (Max)Moonshot	76.7	1,658	—	—	—
49	Muse Wonder (Max)Meta	75.5	1,649	—	—	—
50	DeepSeek v3.2 ReasonerDeepSeek	75.2	—	—	—	60
51	Inkling 2 SmallThinking Machines	75.2	1,646	—	—	—
52	Gemini 3.8 FlashGoogle	75.0	—	73.3	—	—
53	Grok 4.7xAI	74.7	—	73.1	—	—
54	Mimo v2.6 Flash小米集团	74.1	1,638	—	—	—
55	Gemini 3 FlashGoogle	73.3	1,439	—	—	75.8
56	Hy4 PreviewTencent	73.3	1,632	—	—	—
57	Claude 3 7 Sonnet 20250219 32k Thinking Tokens (Thinking)Anthropic	72.6	—	—	64.9	—
58	Claude Fable 5 (High)Anthropic	72.5	1,625	—	—	—
59	qwen3.8 Flash NextAlibaba	72.2	1,624	—	—	—
60	GLM 5.3 (Max)Zhipu	72.1	1,622	—	—	—
61	DeepSeek v4.1 Flash 20260910 (Max)DeepSeek	71.8	1,621	—	—	—
62	DeepSeek R1 + Claude 3 5 Sonnet 20241022DeepSeek	71.6	—	—	64	—
63	GPT 6 LunaOpenAI	71.4	—	70.8	—	—
64	GPT 5.6 Sol Xhigh Codex HarnessOpenAI	71.1	1,615	—	—	—
65	GLM 5.3 FlashZhipu	70.9	1,609	70.9	—	—
66	GLM 5Z-AI	70.8	1,435	—	—	72.8
67	U2 Flashunisound	70.5	1,611	—	—	—
68	Muse Ember (Max)Meta	70.3	1,609	—	—	—
69	GLM 5.2 (Max)zai	70.2	1,608	—	—	—
70	Devstral Small 2512Mistral	69.9	—	—	—	56.4
71	GPT 5.2OpenAI	69.8	1,420	—	—	72.8
72	MiniMax m2.5MiniMax	69.8	1,386	—	—	75.8
73	O1 2024 12 17 (High)OpenAI	68.8	—	—	61.7	—
74	Gemini 3 ProGoogle	68.8	1,441	—	—	69.6
75	qwen3.8 27bAlibaba	68.6	1,590	69.3	—	—
76	Claude Sonnet 4 20250514 32k Thinking (Thinking)Anthropic	68.4	—	—	61.3	—
77	qwen3.8 (Max)Alibaba	67.9	1,670	60.9	—	—
78	MiniMax m3.1 Flash稀宇科技	67.9	1,585	68.8	—	—
79	Gemini 3.7 Flash (High)Google	67.6	1,588	—	—	—
80	Mimo v2.6 Pro小米集团	67.4	1,624	64.5	—	—
81	O3 Mini (High)OpenAI	67.3	—	—	60.4	—
82	Claude 3 7 Sonnet 20250219 No Thinking (Thinking)Anthropic	67.3	—	—	60.4	—
83	Step 5 Preview阶跃星辰	66.9	—	67.6	—	—
84	DeepSeek V4 Flash (High)DeepSeek	66.7	1,582	—	—	—
85	DeepSeek V4 Pro High 20260813DeepSeek	66.7	1,581	—	—	—
86	GLM 4.5Z.ai	66.7	—	—	—	54.2
87	Qwen3 235b A22b Diff No Think Alibaba ApiAlibaba	66.4	—	—	59.6	—
88	Devstral 2512Mistral	66.1	—	—	—	53.8
89	Kimi K2Moonshot	65.8	—	—	59.1	—
90	Claude 3.7 SonnetAnthropic	64.6	—	—	—	52.8
91	GPT 5.2 CodexOpenAI	64.5	1,338	—	—	72.8
92	Claude Opus 4.7 (High)Anthropic	63.5	1,557	—	—	—
93	Claude Opus 4.8 (High)Anthropic	63.3	1,555	—	—	—
94	DeepSeek R1DeepSeek	63.2	—	—	56.9	—
95	Grok 4.5xAI	62.9	1,552	—	—	—
96	Claude Sonnet 4 20250514 No Thinking (Thinking)Anthropic	62.6	—	—	56.4	—
97	Claude Opus 4.6 (High)Anthropic	62.1	1,547	—	—	—
98	DeepSeek v3.2DeepSeek	61.6	1,326	—	—	70
99	Muse Spark 1.1Meta	61.5	1,541	—	—	—
100	Claude Sonnet 5 (High)Anthropic	61.1	1,539	—	—	—
101	DeepSeek V3 0324DeepSeek	61.0	—	—	55.1	—
102	Gemini 2.5 Flash Preview 05 20 24k Think (Thinking)Google	61.0	—	—	55.1	—
103	Gemini 3.6 Flash (High)Google	61.0	1,538	—	—	—
104	Quasar AlphaOther	60.5	—	—	54.7	—
105	Muse Spark 1.2 (xHigh)Meta	60.4	1,534	—	—	—
106	GPT 5.1OpenAI	59.7	1,342	—	—	66
107	Muse Spark Open (Max)Meta	59.5	1,526	—	—	—
108	GPT 5.1 CodexOpenAI	59.5	1,338	—	—	66
109	Claude Sonnet 4.6Anthropic	58.9	1,522	—	—	—
110	Grok 3 BetaxAI	58.9	—	—	53.3	—
111	GPT 5.6 Terra Xhigh Codex HarnessOpenAI	58.7	1,520	—	—	—
112	GPT 5.6 Luna Xhigh Codex HarnessOpenAI	58.5	1,519	—	—	—
113	seed2.1 Pro PreviewByteDance	58.5	1,519	—	—	—
114	Optimus AlphaOther	58.4	—	—	52.9	—
115	qwen3.7 Max 20260517Alibaba	58.0	1,515	—	—	—
116	GPT 5.5 Xhigh Codex HarnessOpenAI	57.6	1,512	—	—	—
117	Kimi k2.6Moonshot	57.3	1,509	—	—	—
118	GLM 5.1zai	57.1	1,508	—	—	—
119	Claude 3 5 Sonnet 20241022Anthropic	56.9	—	—	51.6	—
120	Hy3Tencent	56.8	1,506	—	—	—
121	Gemini 3.5 Flash (High)Google	56.0	1,500	—	—	—
122	Claude Opus 4.5 High 32k 20251101Anthropic	55.1	1,493	—	—	—
123	GPT 5.5 High Codex HarnessOpenAI	54.4	1,487	—	—	—
124	Grok 3 Mini Beta (High)xAI	54.1	—	—	49.3	—
125	Spark x2.6科大讯飞	54.0	1,485	—	—	—
126	MiniMax M3稀宇科技	53.9	1,484	—	—	—
127	qwen3.6 Max PreviewAlibaba	53.7	1,482	—	—	—
128	Mimo v2.5 Proxiaomi	53.3	1,479	—	—	—
129	MiniMax M2MiniMax	53.2	1,299	—	—	61
130	O4 (Mini)OpenAI	53.1	—	—	—	45
131	DeepSeek Chat V3 PrevDeepSeek	53.1	—	—	48.4	—
132	Kimi k2.7 CodeMoonshot	52.7	1,474	—	—	—
133	Claude Opus 4.5 20251101Anthropic	51.9	1,469	—	—	—
134	GLM 4.6Z.ai	51.9	1,342	—	—	55.4
135	Gemini 2.5 Flash Preview 04 17 (Default) Google	51.5	—	—	47.1	—
136	GPT 4.1OpenAI	51.5	—	—	52.4	39.6
137	Kimi K2 InstructMoonshot	51.3	—	—	—	43.8
138	DeepSeek V4 Pro Preview (High)DeepSeek	51.2	1,463	—	—	—
139	GPT 5.4 High Codex HarnessOpenAI	51.1	1,462	—	—	—
140	qwen3.6 PlusAlibaba	51.0	1,461	—	—	—
141	GPT 5.5 Codex HarnessOpenAI	50.2	1,455	—	—	—
142	ChatGPT 4o Latest 2025 03 29OpenAI	49.4	—	—	45.3	—
143	DeepSeek V4 ProDeepSeek	49.3	1,448	—	—	—
144	Mistral Large 4Mistral	49.3	1,564	44.4	—	—
145	Gemini 3.1 Pro PreviewGoogle	49.0	1,446	—	—	—
146	GPT 4.5 PreviewOpenAI	48.9	—	—	44.9	—
147	GPT 5.4OpenAI	48.5	1,442	—	—	—
148	Mimo v2.5xiaomi	47.9	1,438	—	—	—
149	Gemini 2.5 Flash Preview 05 20 No Think (Thinking)Google	47.9	—	—	44	—
150	Nemotron Super Vl 3.5NVIDIA	47.8	—	54.1	—	—
151	Beam 1.0 501b A23breflection	47.7	1,437	—	—	—
152	Gemini 3.5 Flash LiteGoogle	47.6	1,435	—	—	—
153	GLM 4.7zai	47.5	1,435	—	—	—
154	Qwen3 Coder 480b A35b InstructAlibaba	47.5	1,275	—	—	55.4
155	Mimo V2 Proxiaomi	47.4	1,434	—	—	—
156	DeepSeek V4 Flash Preview (High)DeepSeek	47.0	1,431	—	—	—
157	GPT Oss 120b (High)OpenAI	45.3	—	—	41.8	—
158	Smaug Nemotron 3.5 Nvfp4NVIDIA	44.1	—	51.5	—	—
159	Inkling SmallThinking Machines	43.9	1,408	—	—	—
160	GPT 5.3 Codex Codex HarnessOpenAI	43.7	1,406	—	—	—
161	Kimi k2.5 InstantMoonshot	43.7	1,405	—	—	—
162	Qwen3 32bAlibaba	43.1	—	—	40	—
163	Nvidia Nemotron 3.5 Super Vl 120b A12b Nvfp4NVIDIA	43.1	1,401	—	—	—
164	Gemini 2.5 ProGoogle	43.1	1,228	—	—	53.6
165	qwen3.5 397b A17bAlibaba	43.0	1,400	—	—	—
166	GLM 5v Turbozai	42.8	1,399	—	—	—
167	MiniMax m2.7MiniMax	42.7	1,398	—	—	—
168	GPT 5.4 Mini (High)OpenAI	42.7	1,398	—	—	—
169	Inklingthinky	42.5	1,414	48.7	—	—
170	Claude Sonnet 4.5 High 32k 20250929Anthropic	42.1	1,393	—	—	—
171	Claude Opus 4.1 20250805Anthropic	41.6	1,389	—	—	—
172	Gemini Exp 1206Google	41.0	—	—	38.2	—
173	Claude Sonnet 4.5 20250929Anthropic	41.0	1,385	—	—	—
174	MiniMax m2.1 PreviewMiniMax	40.9	1,384	—	—	—
175	Gemini 3 Flash MinimalGoogle	40.9	1,384	—	—	—
176	Grok 4.20 Beta Reasoning 0309xAI	39.7	1,375	—	—	—
177	Solar Pro 4Upstage	38.9	1,369	—	—	—
178	GPT 5 NanoOpenAI	38.1	—	—	—	34.8
179	Gemini 2.0 Pro Exp 02 05Google	37.9	—	—	35.6	—
180	qwen3.5 122b A10bAlibaba	37.7	1,360	—	—	—
181	Gemma 4 26b A4bGoogle	37.5	1,358	—	—	—
182	Hy3 PreviewTencent	37.5	1,358	—	—	—
183	Grok 4.3xAI	37.4	1,357	—	—	—
184	qwen3.5 27bAlibaba	37.4	1,357	—	—	—
185	Muse GlimmerMeta	37.0	1,355	—	—	—
186	Laguna m.1poolside	36.4	1,350	—	—	—
187	O1 Mini 2024 09 12OpenAI	34.7	—	—	32.9	—
188	Mimo V2 Flash Non Thinking (Thinking)xiaomi	33.9	1,331	—	—	—
189	Claude Haiku 4.5 20251001Anthropic	33.8	1,330	—	—	—
190	Kimi K2 Thinking TurboMoonshot	32.7	1,322	—	—	—
191	Laguna xs.2poolside	30.3	1,303	—	—	—
192	Gemini 2.5 FlashGoogle	29.1	—	—	—	28.7
193	Claude 3 5 Haiku 20241022Anthropic	28.9	—	—	28	—
194	Mimo V2 Flash (Thinking)xiaomi	28.8	1,292	—	—	—
195	ChatGPT 4o Latest 2025 02 15OpenAI	27.8	—	—	27.1	—
196	Qwq 32b + Qwen 2.5 Coder InstructOther	26.8	—	—	26.2	—
197	DeepSeek v3.2 ExpDeepSeek	26.0	1,270	—	—	—
198	Mistral Medium 3.5Mistral	25.0	1,263	—	—	—
199	Kat Coder Pro V1Other	24.3	1,258	—	—	—
200	Gemini 3.1 Flash Lite PreviewGoogle	24.0	1,256	—	—	—
201	qwen3.5 35b A3bAlibaba	23.7	1,253	—	—	—
202	GPT 4o 2024 08 06OpenAI	23.1	—	—	23.1	—
203	qwen3.5 FlashAlibaba	22.5	1,244	—	—	—
204	Grok 4.1 Fast ReasoningxAI	22.2	1,242	—	—	—
205	Gemini 2.0 Flash ExpGoogle	22.0	—	—	22.2	—
206	Qwen Max 2025 01 25Alibaba	21.6	—	—	21.8	—
207	Trinity Large ThinkingArcee AI	21.5	1,236	—	—	—
208	Mistral Large 3Mistral	21.1	1,233	—	—	—
209	Nemotron Super Vl 3.5 Nvfp4NVIDIA	20.5	—	34.8	—	—
210	Qwq 32bAlibaba	20.5	—	—	20.9	—
211	Gemma 4 31bGoogle	19.5	1,370	20.3	—	—
212	Grok 4.1 (Thinking)xAI	19.0	1,217	—	—	—
213	GPT 4oOpenAI	18.6	—	—	—	21.6
214	Llama 4 Maverick InstructMeta	17.8	—	—	—	21.0
215	GPT 4o 2024 11 20OpenAI	17.3	—	—	18.2	—
216	Gemini 2.0 Flash Thinking Exp 01 21Google	17.3	—	—	18.2	—
217	DeepSeek Chat v2.5DeepSeek	16.8	—	—	17.8	—
218	Devstral 2Mistral	16.2	1,196	—	—	—
219	Granite 4.1 8bIBM	15.8	1,193	—	—	—
220	Llama 4 MaverickMeta	14.2	—	—	15.6	—
221	Mercury 2inception-ai	12.9	1,171	—	—	—
222	Grok Code Fast 1xAI	12.1	1,164	—	—	—
223	Grok 4 Fast ReasoningxAI	11.8	1,162	—	—	—
224	Yi LightningOther	11.0	—	—	12.9	—
225	Command A 03 2025 QualityOther	10.0	—	—	12	—
226	Codestral 25.01Mistral	8.9	—	—	11.1	—
227	Openhands Lm 32b v0.1Other	7.8	—	—	10.2	—
228	qwen2.5 Coder 32b InstructAlibaba	7.6	—	—	16.4	9
229	Gemini 2.0 FlashGoogle	6.7	—	—	—	13.5
230	GPT 4.1 NanoOpenAI	6.3	—	—	8.9	—
231	Gemma 3 27b ItGoogle	1.5	—	—	4.9	—
232	Llama 4 Scout InstructMeta	0.1	—	—	—	9.1
233	GPT 4o Mini 2024 07 18OpenAI	0	—	—	3.6	—
234	Devstral Medium 2507Mistral	0	1,072	—	—	—

Sources

LMArena WebDev Arena — Official webdev / latest snapshot, measuring front-end and web-app battles.
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.

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.