Arena — Code

按 Arena 自己发布的样子呈现——他们的指标，他们的模型。点击模型名可前往其上游页面。

发布方: Arena
许可: 未声明
更新频率: 持续更新
数据截至: 2026-08-21
抓取时间: 2026-08-22
条目数: 118 · 23 条未匹配

按 Overall 排序的前 12 名，每个模型一行

Claude Opus 5 1691
Kimi K3 1674
qwen3.8-max 1669
Grok 4.6 1629
Claude Fable 5 (Opus 4.8 fallback) 1626
GPT-5.6 Sol 1619
GLM-5.3 1599
qwen3.8-27b 1595
Gemini 3.7 Flash 1587
GLM-5.2 1582
deepseek-v4-pro-high-20260813 1582
deepseek-v4-flash-high 1579

Elo 没有零点——它是一个任意的锚，而不是“毫无能力”——所以从它画起的条形会把这个锚一并编码进去。此处每一行是画在一条起点为 1566 的范围上的点；可读的是行与行之间的距离，而不是它到左边缘的距离。

排序依据仅显示已匹配的模型

118 个有分数

#	模型
1	Claude Opus 5maxanthropic	16911682–1699
2	Kimi K3maxkimi	16741662–1685
3	qwen3.8-maxmax未匹配	16691656–1681
4	Claude Opus 5highanthropic	16631655–1671
5	Grok 4.6highxai	16291612–1646
6	Claude Fable 5 (Opus 4.8 fallback)anthropic	16261618–1634
7	GPT-5.6 Solcodex-harness · xhighopenai	16191611–1626
8	GLM-5.3maxzai	15991583–1614
9	qwen3.8-27b未匹配	15951582–1608
10	Gemini 3.7 Flashhighgoogle	15871574–1600
11	GLM-5.2maxzai	15821574–1590
12	deepseek-v4-pro-high-20260813未匹配	15821570–1594
13	deepseek-v4-flash-highhigh未匹配	15791568–1590
14	Claude Opus 4.8highanthropic	15631556–1571
15	Claude Opus 4.7anthropic	15581551–1564
16	Claude Opus 4.7highanthropic	15571550–1563
17	Grok 4.5xai	15561548–1565
18	Claude Opus 4.6highanthropic	15461540–1551
19	Claude Opus 4.8anthropic	15391532–1546
20	Muse Spark 1.1meta	15391531–1548
21	Gemini 3.6 Flashhighgoogle	15391530–1547
22	Claude Sonnet 5highanthropic	15391531–1547
23	Claude Opus 4.6anthropic	15361531–1542
24	Muse Spark 1.2xhighmeta	15341520–1548
25	Claude Sonnet 4.6anthropic	15221517–1528
26	seed-2.1-pro-preview未匹配	15211514–1529
27	GPT-5.6 Terracodex-harness · xhighopenai	15201511–1529
28	Hy3tencent	15181506–1531
29	GPT-5.6 Lunacodex-harness · xhighopenai	15181509–1526
30	qwen3.7-max-20260517未匹配	15171509–1524
31	GLM-5.1zai	15091502–1516
32	Kimi K2.6kimi	15091501–1516
33	GPT-5.5codex-harness · xhighopenai	15081502–1515
34	Gemini 3.5 Flashhighgoogle	14991491–1507
35	claude-opus-4-5-20251101-high-32k未匹配	14941486–1503
36	Gemini 3.5 Flashmediumgoogle	14901483–1497
37	MiniMax-M3minimax	14881481–1495
38	GPT-5.5codex-harness · highopenai	14861480–1492
39	qwen3.6-max-preview未匹配	14791466–1492
40	MiMo-V2.5-Proxiaomi	14761469–1482
41	Kimi K2.7 Codekimi	14731463–1482
42	claude-opus-4-5-20251101未匹配	14681461–1475
43	deepseek-v4-pro-high-preview未匹配	14641457–1470
44	GPT-5.4codex-harness · highopenai	14631443–1482
45	Qwen3.6 Plusalibaba	14601454–1466
46	GPT-5.5codex-harnessopenai	14581451–1464
47	Gemini 3.5 Flash-Litegoogle	14491407–1492
48	Gemini 3.1 Pro Previewgoogle	14461441–1452
49	DeepSeek V4 Pro 0813deepseek	14451439–1452
50	GPT-5.4codex-harness · mediumopenai	14421423–1461
51	Gemini 3 Flash Previewgoogle	14381429–1447
52	Gemini 3 Pro Previewgoogle	14381429–1446
53	MiMo-V2.5xiaomi	14381431–1444
54	Kimi K2.5onkimi	14361431–1441
55	GLM-5zai	14351427–1444
56	GLM-4.7zai	14341422–1447
57	MiMo-V2-Proxiaomi	14341425–1442
58	deepseek-v4-flash-high-preview未匹配	14311424–1438
59	GPT-5mediumopenai	14191403–1435
60	GPT-5.2openai	14181395–1441
61	GPT-5.3 Codexcodex-harnessopenai	14081395–1422
62	Inklingthinking-machines	14081400–1416
63	kimi-k2.5-instant未匹配	14051393–1417
64	Inkling Smallthinking-machines	14021391–1412
65	GLM 5V Turbozai	14001386–1414
66	Qwen3.5 397B A17Balibaba	13991394–1405
67	MiniMax-M2.7minimax	13981392–1404
68	GPT-5.4 minihighopenai	13971390–1404
69	claude-sonnet-4-5-20250929-high-32k未匹配	13921385–1400
70	GPT-5.1mediumopenai	13911379–1403
71	GPT-5.4openai	13911376–1405
72	claude-opus-4-1-20250805未匹配	13891378–1400
73	minimax-m2.1-preview未匹配	13871377–1397
74	claude-sonnet-4-5-20250929未匹配	13861378–1393
75	MiniMax-M2.5minimax	13841375–1392
76	Gemini 3 Flash Previewon · minimalgoogle	13831377–1388
77	grok-4.20-beta-0309-reasoningon未匹配	13741367–1380
78	Solar Pro 4upstage	13711354–1388
79	GPT-5.3 Codexcodex-harnessopenai	13691357–1381
80	Gemma 4 31Bgoogle	13641356–1371
81	Gemma 4 26B A4Bgoogle	13621344–1379
82	DeepSeek V3.2ondeepseek	13601351–1370
83	Muse Glimmermeta	13591343–1375
84	Qwen3.5 122B A10Balibaba	13581350–1365
85	Qwen3.5 27Balibaba	13571349–1366
86	Grok 4.3xai	13571350–1363
87	Hy3-previewtencent	13561339–1374
88	laguna-m.1未匹配	13471338–1357
89	GPT-5.1openai	13411332–1349
90	GLM-4.6zai	13401329–1351
91	GPT-5.2 Codexopenai	13381329–1347
92	GPT-5.1 Codexopenai	13361325–1348
93	MiMo-V2-Flashoffxiaomi	13301320–1340
94	claude-haiku-4-5-20251001未匹配	13271322–1332
95	DeepSeek V3.2deepseek	13251316–1333
96	Kimi K2 Thinkingkimi	13231315–1330
97	laguna-xs.2未匹配	13021291–1313
98	MiniMax-M2minimax	12971286–1308
99	MiMo-V2-Flashonxiaomi	12921276–1309
100	Qwen3 Coder 480B A35B Instructalibaba	12731265–1281
101	DeepSeek V3.2 Expdeepseek	12721258–1285
102	Mistral Medium 3.5mistral	12651250–1280
103	KAT-Coder-Pro V1kwaikat	12551236–1275
104	Gemini 3.1 Flash-Litegoogle	12541247–1261
105	Qwen3.5 35B A3Balibaba	12501232–1268
106	GPT-5.1 Codex miniopenai	12441221–1266
107	Grok 4.1 Fastonxai	12401229–1251
108	Trinity Large Thinkingarcee	12381218–1259
109	qwen3.5-flash未匹配	12381218–1258
110	Mistral Large 3mistral	12301204–1256
111	Gemini 2.5 Progoogle	12261210–1242
112	grok-4.1-thinkingon未匹配	12101185–1235
113	Devstral 2mistral	11941172–1215
114	Granite 4.1 8Bibm	11911172–1210
115	Mercury 2inception	11661141–1191
116	Grok Code Fast 1xai	11651136–1193
117	Grok 4 Fastonxai	11621134–1190
118	devstral-medium-2507未匹配	10801049–1111

各列衡量的是什么

Overall: 越高越好 · 编程