LMArena 人类偏好榜

名次与分数均取自官方榜单，与官方一致。条形长度按当前这些成绩里的最大值算，领跑那条填满整列，其余按比例缩短（灰斜纹是「额外真实缺陷」，只与同列比、不与分数比）。官方没给的数据显示「—」。

LMArena 人类偏好榜 Elo 分排名

名次	模型	投票数	95% 区间（±）	Elo 分
1	Gemini 4 Argon (High)	10,083	6.6	1526
2	Claude Opus 5.5 (High)	10,322	6.8	1516
3	Claude Fable 5.1 (Max)	16,343	5.9	1513
4	Claude Opus 5 (Max)	34,526	4.7	1508
5	Claude Opus 5 (High)	71,020	3.6	1503
6	Claude Opus 4.6 (High)	83,454	3.3	1501
7	Muse Wonder (Max)	6,001	8.1	1501
8	Gemini 3.8 Flash (High)	43,997	4.5	1498
9	Claude Opus 4.6	88,206	3.2	1497
10	Muse Spark 1.3 (Max)	20,506	5.3	1491
11	MiMo-V2.6-Pro	4,067	9.3	1491
12	Claude Fable 5 (High)	44,253	4.2	1491
13	Claude Opus 4.7 (High)	65,140	3.7	1490
14	Gemini 3.7 Flash (High)	30,501	4.7	1490
15	Qwen3.8-Max	30,500	4.8	1484
16	Claude Opus 4.7	66,183	3.7	1483
17	Muse Spark 1.2 (xHigh)	10,470	6.6	1482
18	Gemini 3.5 Flash (High)	48,504	3.9	1482
19	Gemini 3.1 Pro Preview	130,819	2.9	1480
20	Gemini 3.6 Flash (High)	36,102	4.4	1480
21	Gemini 3 Pro	42,054	3.8	1479
22	Muse Spark 1.1	46,121	4.1	1478
23	Gemini 3.5 Flash (Medium)	47,271	4.1	1478
24	Qwen3.7-Max-Preview	3,941	9.7	1475
25	Kimi K3 (Max)	35,379	4.5	1475
26	Muse Spark	14,147	5.7	1474
27	GLM-5.3 (Max)	25,144	5	1474
28	Claude Sonnet 5.5 (xHigh)	8,977	7	1473
29	GLM-5.3-Flash	38,359	4.7	1472
30	GPT-5.5 (High)	69,375	3.7	1472
31	Qwen3.5-Max-Preview	22,601	4.9	1470
32	GLM-5.2 (Max)	52,283	4.1	1470
33	GPT-5.4 (High)	64,757	3.7	1469
34	ERNIE 5.1	39,474	4.5	1468
35	GPT-5.5	70,998	3.7	1467
36	Gemini 3 Flash	31,313	4.3	1466
37	DeepSeek-V4.1-Flash-20260910 (Max)	13,750	6	1464
38	MiMo-V2.5-Pro	77,883	3.5	1464
39	GLM-5.1	65,402	3.5	1463
40	Claude Opus 4.8 (High)	64,096	3.9	1461
41	Claude Sonnet 4.6	70,953	3.5	1458
42	Gemini 2.5 Pro	125,316	2.4	1458
43	GPT-5.6 Sol (xHigh)	42,713	4.3	1457
44	Kimi K2.6	40,026	4.4	1455
45	Qwen3.7-Plus	42,058	4.3	1455
46	Claude Opus 4.8	65,208	3.8	1454
47	GPT-5.4	67,855	3.7	1452
48	MiMo-V2.6-Flash	13,730	6	1452
49	DeepSeek-V4-Pro	57,588	3.9	1451
50	Claude Opus 4.5 (20251101)	72,850	3.1	1451
51	Grok 4.20 Beta (Reasoning, 0309)	66,418	3.7	1451
52	Grok 4.20 Multi-Agent Beta (0309)	64,928	3.7	1451
53	amazon-nova-experimental-chat (26-02-10)	3,522	9.6	1449
54	Dola Seed 2.0 Pro	79,130	3.3	1448
55	GPT-5.6 Terra (xHigh)	46,470	4.1	1448
56	Claude Opus 4.5 (High, 32K, 20251101)	37,485	3.8	1448
57	Grok 4.5	39,880	4.3	1448
58	Qwen3.6-Max-Preview	5,423	8.2	1447
59	GLM-5	29,267	4.2	1446
60	DeepSeek-V4-Pro (High, 20260813)	9,797	6.6	1446
61	Kimi K2.5 (Thinking)	74,744	3.3	1445
62	DeepSeek-V4-Pro Preview (High)	54,896	3.9	1445
63	NVIDIA-Nemotron-3-Ultra-550B-A55B (NVFP4)	11,557	6.9	1445
64	ERNIE 5.0 (0110)	36,481	3.9	1445
65	Grok 4.20 Beta 1	27,914	4.5	1445
66	GPT-6.1 Sol (Max)	8,305	7.3	1444
67	Gemma 4 31B	6,137	7.4	1443
68	GPT-5.1 (High)	41,646	3.7	1443
69	ERNIE-5.0-Preview (1203)	9,805	6.5	1443
70	Gemini 3 Flash (Minimal)	90,998	3	1442
71	Claude Sonnet 5 (High)	51,784	4	1442
72	Inkling	43,705	4.2	1442
73	GPT-6 Astra (Max)	14,076	6	1442
74	Step 5 Preview	9,792	6.6	1442
75	GLM-4.6	35,818	3.9	1440
76	Qwen3.8-27B	26,238	4.9	1440
77	Qwen3-Max-Preview	27,367	4.5	1439
78	GPT-5.2 Chat (20260210)	35,982	4	1438
79	Claude Sonnet 4.5 (20250929)	82,650	2.9	1438
80	Hy3	13,220	6	1438
81	Qwen3.5-397B-A17B	96,206	3	1438
82	GLM-5V-Turbo	10,477	6.4	1438
83	Qwen3.6-Plus	48,142	4.1	1436
84	Grok 4.1 (Thinking)	66,479	3.1	1436
85	MiMo-V2-Pro	25,435	4.6	1436
86	Grok 4.1	68,621	3.1	1436
87	GLM-4.7	12,245	6	1435
88	Gemini 3.5 Flash-Lite	35,829	4.5	1435
89	Claude Sonnet 4.5 (High, 32K, 20250929)	83,906	2.8	1435
90	Gemma 4 26B A4B	6,092	7.4	1434
91	DeepSeek-V4-Flash	51,982	4	1432
92	Mistral Large 4	10,331	6.4	1432
93	MiniMax-M3	67,030	3.7	1432
94	GPT-5.6 Luna (xHigh)	41,426	4.3	1431
95	GLM-4.5	23,791	4.9	1430
96	ERNIE-5.0-Preview (1022)	4,754	8.8	1430
97	Grok 4.6 (High)	32,189	4.7	1430
98	ChatGPT-4o Latest (20250326)	81,905	2.8	1428
99	Mistral Large 3	86,928	2.8	1428
100	MiMo-V2.5	47,364	4.2	1428
101	DeepSeek-R1 (0528)	18,170	5.7	1428
102	LongCat-Flash-Chat Exp (2602)	29,275	4.5	1427
103	Beam 1.0 501B A23B	4,237	9.3	1426
104	Grok 3 Preview (02-24)	32,498	4.3	1425
105	DeepSeek-V3.2-Exp (Thinking)	8,994	6.6	1425
106	DeepSeek-V3.2	48,090	3.5	1425
107	DeepSeek-V4-Flash Preview (High)	51,781	4.1	1424
108	Mistral Medium 3.1 (2508)	95,285	2.6	1424
109	MiMo-V2-Omni	20,618	5.5	1423
110	LongCat-Flash-Chat	11,166	6.4	1423
111	GPT-5.1	44,419	3.6	1422
112	DeepSeek-V3.2-Exp	11,997	6.3	1422
113	Mistral Medium 3.5	11,791	6.4	1421
114	Kimi K2.5 (Instant)	8,371	6.5	1420
115	Qwen3-VL-235B-A22B-Instruct	11,312	6.5	1420
116	DeepSeek-V3.2 (Thinking)	42,042	3.6	1420
117	DeepSeek-V3.1	14,607	6	1420
118	amazon-nova-experimental-chat (12-10)	3,728	9.4	1420
119	DeepSeek-V3.1-Terminus (Thinking)	3,375	10.1	1419
120	GPT-5.5 Instant	27,163	4.9	1419
121	Claude Opus 4.1 (Thinking 16K, 20250805)	49,536	3.5	1419
122	Qwen3-235B-A22B-Instruct (2507)	98,096	2.5	1419
123	Claude Opus 4.1 (20250805)	77,211	3	1419
124	Qwen3.5-122B-A10B	29,716	4.3	1418
125	GPT-4.5 Preview (2025-02-27)	14,547	5.7	1417
126	DeepSeek-V3.1-Terminus	3,608	9.7	1417
127	Qwen3-Next-80B-A3B-Instruct	22,755	4.7	1417
128	Gemini 2.5 Flash	125,462	2.4	1417
129	DeepSeek-V3.1 (Thinking)	11,454	6.6	1417
130	GPT-5.2 (High)	49,643	3.5	1416
131	Gemini 3.1 Flash-Lite Preview	64,209	3.6	1416
132	amazon-nova-experimental-chat (11-10)	26,031	4.2	1415
133	Qwen3-235B-A22B-Thinking (2507)	8,821	6.5	1415
134	Kimi K2 Thinking Turbo	63,348	3.1	1414
135	Inkling-Small	35,359	4.5	1413
136	Qwen3-Max (2025-09-23)	9,027	6.5	1413
137	GPT-5.2	83,828	3.1	1412
138	GPT-5.4 mini (High)	63,451	3.7	1412
139	Grok 4 (0709)	39,679	3.9	1411
140	MiMo-V2-Flash (Non-Thinking)	48,069	3.4	1411
141	o3 (2025-04-16)	58,894	3.6	1410
142	Qwen3.5-27B	28,661	4.3	1409
143	Grok 4.1 Fast (Reasoning)	57,526	3.2	1408
144	Grok 4 Fast (Non-Reasoning)	6,337	7.8	1407
145	Gemini 2.5 Flash Preview (09-2025)	32,982	4	1406
146	GPT-5 (High)	31,712	4.4	1406
147	Hunyuan Vision 1.5 Thinking	2,168	12.4	1406
148	NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)	4,176	9.3	1405
149	MiniMax-M2.7	87,731	3.3	1405
150	GPT-5 Chat	31,183	4.3	1403
151	Step-3.5-Flash	60,329	3.5	1403
152	Hy3 preview	6,875	7.4	1401
153	Claude Haiku 5.5 (High)	6,080	8.3	1401
154	Qwen3-VL-235B-A22B-Thinking	7,855	6.8	1401
155	Hunyuan T1 (20250711)	4,595	8.6	1401
156	Grok 4 Fast (Reasoning)	18,267	5	1400
157	GPT-6 Sol (Max)	14,707	6	1398
158	Grok 4.7 (xHigh)	14,693	5.9	1398
159	Qwen3.5-Flash	61,802	3.6	1397
160	Grok 4.3	71,348	3.6	1397
161	Claude Haiku 4.5 (20251001)	152,393	2.4	1397
162	amazon-nova-experimental-chat (26-01-10)	3,537	9.6	1396
163	MiMo-V2-Flash (Thinking)	11,140	6.1	1395
164	Qwen3.5-35B-A3B	30,443	4.2	1395
165	amazon-nova-experimental-chat (10-20)	11,621	6.1	1394
166	Qwen3-235B-A22B (Non-Thinking)	37,485	4.5	1393
167	MiniMax-M2.1 Preview	17,230	5.1	1392
168	GPT-6 Luna (Max)	72,504	4.3	1390
169	Muse Glimmer	3,889	9.7	1389
170	GPT-5.3 Chat	34,324	4.2	1389
171	Solar Pro 4	24,453	5.1	1388
172	Qwen3-30B-A3B-Instruct (2507)	23,306	4.9	1384
173	GLM-4.5-Air	30,646	4.2	1384
174	GPT-4.1 (2025-04-14)	50,302	3.7	1383
175	Kimi K2 Preview (0905)	11,586	6.5	1380
176	Gemini 2.5 Flash-Lite Preview (Non-Thinking, 09-2025)	47,420	3.4	1379
177	NVIDIA-Nemotron-3-Super-120B-A12B	7,869	7.1	1378
178	Hunyuan TurboS (20250416)	10,623	6.4	1377
179	Claude Opus 4 (Thinking 16K, 20250514)	35,908	4.4	1377
180	GLM-4.6V	2,831	11.1	1376
181	DeepSeek-V3 (0324)	44,974	3.9	1375
182	GPT-5 mini (High)	26,799	4.6	1373
183	DeepSeek-R1	18,524	4.8	1373
184	GPT-5.4 nano (High)	62,375	3.8	1372
185	Kimi K2 Preview (0711)	27,136	4.9	1371
186	Mistral Medium 3 (2505)	32,776	4.7	1370
187	Grok 3 Mini (High)	15,591	5.4	1369
188	Gemini 2.5 Flash-Lite Preview (Thinking, 06-17)	32,380	4.5	1369
189	Qwen3-Next-80B-A3B-Thinking	13,473	5.8	1368
190	Qwen2.5-Max	32,487	4.1	1367
191	Qwen3-235B-A22B	25,949	4.7	1366
192	Claude Opus 4 (20250514)	43,036	4.3	1366
193	o1 (2024-12-17)	27,807	4.5	1366
194	gpt-oss-120b	30,207	4.3	1366
195	Grok 3 Mini Beta	20,929	5.1	1364
196	amazon-nova-experimental-chat (10-09)	2,781	10.9	1364
197	Amazon Nova 2 Lite	12,410	6	1362
198	Ling-flash-2.0	6,953	7.2	1361
199	Granite 4.2 30B	3,407	10.4	1361
200	MiniMax-M2.5	42,791	3.9	1359
201	Gemma 3 27B IT	46,665	3.7	1358
202	Qwen3-Coder-480B-A35B-Instruct	25,192	5	1357
203	INTELLECT-3	5,432	8.2	1356
204	Mercury 2	3,193	10.4	1355
205	Gemini 2.0 Flash (001)	43,443	3.7	1354
206	o4-mini (2025-04-16)	44,851	4	1353
207	o1-preview	31,122	5	1353
208	Claude Sonnet 4 (Thinking 32K, 20250514)	34,135	4.5	1351
209	GLM-4.7-Flash	12,013	5.7	1351
210	Step-3	6,295	7.4	1350
211	NVIDIA-Nemotron-3-Nano-30B-A3B (BF16)	15,765	5.4	1348
212	MiniMax-M1	34,299	4.3	1343
213	GPT-4.1 mini (2025-04-14)	38,701	4.3	1341
214	MiniMax-M2	6,855	7.7	1340
215	Trinity Large Thinking	30,307	4.6	1340
216	Qwen3-32B	3,926	9.3	1340
217	Claude Sonnet 4 (20250514)	39,167	4.4	1339
218	Trinity Large Preview	31,246	4.2	1339
219	Llama 3.3 Nemotron Super 49B v1.5	3,328	10	1338
220	Mistral Small 3.2 (2506)	17,454	5.2	1338
221	o3-mini (High)	18,589	5.2	1337
222	Step-1o Turbo (202506)	8,904	6.9	1336
223	Gemma 3 12B IT	3,829	9.4	1334
224	Ring-flash-2.0	6,987	7.2	1334
225	GLM-4.5V	4,821	8.4	1333
226	DeepSeek-V3	21,770	4.7	1332
227	Command A (03-2025)	55,758	3.4	1331
228	GLM-4-Plus (0111)	5,760	8.4	1331
229	NVIDIA-Nemotron-3.5-Lightning-30B-A3B (NVFP4)	22,498	5.4	1331
230	Gemini 2.0 Flash-Lite Preview (02-05)	24,955	4.3	1330
231	QwQ-32B	25,174	4.4	1329
232	Qwen-Plus (0125)	5,819	8.3	1326
233	Step-2 16K Exp (202412)	4,833	8.5	1321
234	GPT-5 nano (High)	8,134	6.8	1320
235	Hunyuan TurboS (20250226)	2,220	11.7	1320
236	Llama 3.1 Nemotron Ultra 253B v1	2,549	11.5	1320
237	o3-mini	56,742	3.5	1319
238	Granite 4.2 8B	3,259	11.1	1319
239	Gemini 1.5 Pro (002)	55,606	3.4	1319
240	o1-mini	51,981	3.6	1317
241	Qwen3-30B-A3B	26,155	4.7	1317
242	Claude 3.7 Sonnet (Thinking 32K, 20250219)	38,158	4.2	1314
243	Hunyuan Turbo (0110)	2,290	11.4	1311
244	Olmo 3.1 32B Instruct	11,805	6.1	1311
245	Llama 3.3 Nemotron Super 49B v1	2,218	12	1308
246	Gemma 3n E4B IT	22,142	5.2	1306
247	Grok 2 (2024-08-13)	63,498	3.7	1305
248	Yi Lightning	27,332	4.9	1302
249	GPT-4o (2024-05-13)	112,881	3.4	1300
250	Olmo 3 32B Think	5,825	8.2	1300
251	Claude 3.7 Sonnet (20250219)	42,572	4	1299
252	Qwen2.5-Plus (1127)	10,187	6.3	1299
253	Claude 3.5 Sonnet (20241022)	87,771	3.2	1298
254	DeepSeek-V2.5 (1210)	6,795	8.2	1294
255	Granite 4.2 3B	3,260	11.5	1294
256	Athene V2 Chat	24,739	4.5	1291
257	Gemma 3 4B IT	4,171	9.2	1291
258	Granite 4.1 8B	4,177	9.9	1291
259	GLM-4-Plus	26,126	4.9	1290
260	Molmo 2 8B	823	20.7	1288
261	Hunyuan Large (2025-02-10)	3,738	9.7	1288
262	gpt-oss-20b	10,417	6.4	1287
263	Llama 4 Maverick 17B 128E Instruct	39,444	4.3	1287
264	Gemini 1.5 Flash (002)	34,902	4.2	1287
265	GPT-4o mini (2024-07-18)	68,705	3.5	1286
266	GPT-4.1 nano (2025-04-14)	6,103	7.6	1285
267	Llama 3.1 405B Instruct (BF16)	41,375	3.7	1284
268	Llama 3.1 Nemotron 70B Instruct	7,140	7.7	1283
269	GPT-4o (2024-08-06)	45,499	4.2	1283
270	Mercury	1,964	13.7	1282
271	Qwen-Max (0919)	16,478	5.7	1282
272	Llama 3.1 405B Instruct (FP8)	59,656	3.6	1282
273	Claude 3.5 Sonnet (20240620)	82,419	3.4	1281
274	Grok 2 mini (2024-08-13)	52,567	3.7	1281
275	Llama 4 Scout 17B 16E Instruct	29,830	4.7	1279
276	Gemini Advanced (0514)	50,148	5.1	1279
277	Mistral Small 3.1 24B Instruct (2503)	32,721	4.5	1277
278	Llama 3.3 70B Instruct	54,493	3.5	1274
279	Hunyuan Standard (2025-02-10)	3,904	9.6	1274
280	Gemini 1.5 Pro (001)	79,138	4	1273
281	GPT-4 Turbo (2024-04-09)	98,114	3.9	1272
282	Olmo 3.1 32B Think	8,299	7.2	1271
283	DeepSeek-V2.5	24,572	4.7	1271
284	Qwen2.5-72B-Instruct	39,406	4.1	1269
285	Mistral Large 2 (2407)	45,459	3.9	1266
286	Mistral Large (2411)	28,073	4.4	1265
287	Athene 70B (0725)	19,621	5.6	1265
288	GPT-4 Turbo Preview (1106)	100,105	3.9	1263
289	Hunyuan Large Vision	5,283	9.2	1263
290	GPT-4 Turbo Preview (0125)	93,439	4.1	1262
291	Claude 3 Opus (20240229)	194,909	3	1262
292	Llama 3.1 70B Instruct	55,240	3.7	1261
293	Amazon Nova Pro 1.0	24,745	4.5	1259
294	Llama 3.1 Tulu 3 70B	2,846	10.5	1256
295	Claude 3.5 Haiku (20241022)	69,055	3.2	1255
296	Magistral Medium (2506)	11,509	6.4	1255
297	Reka Core (20240904)	7,312	7.1	1248
298	Granite-4.0-H-Small	5,628	8.4	1242
299	Gemini 1.5 Flash (001)	62,833	4.5	1239
300	Jamba 1.5 Large	8,662	7.3	1237
301	Mistral Small 24B Instruct (2501)	14,681	5.9	1234
302	Gemma 2 27B IT	75,754	3.4	1231
303	Qwen2.5-Coder-32B-Instruct	5,432	8	1230
304	Command R+ (08-2024)	9,866	6.5	1229
305	Amazon Nova Lite 1.0	19,372	5.2	1228
306	Llama 3.1 Nemotron 51B Instruct	3,749	9.9	1228
307	Gemma 2 9B IT SimPO	10,072	6.8	1227
308	GLM-4 (0520)	9,788	6.9	1226
309	Gemini 1.5 Flash-8B (001)	35,558	4.3	1226
310	Nemotron-4 340B Instruct	19,659	5.3	1225
311	Aya Expanse 32B	27,124	4.9	1224
312	Llama 3 70B Instruct	156,876	3.6	1221
313	Claude 3 Sonnet (20240229)	109,284	4	1218
314	OLMo 2 32B Instruct (0325)	3,334	10.8	1218
315	Reka Flash (20240904)	7,536	6.9	1218
316	Phi-4	24,126	4.6	1217
317	Amazon Nova Micro 1.0	19,364	5.2	1208
318	Gemma 2 9B IT	54,611	3.8	1207
319	GPT-4 (0314)	54,173	4.7	1206
320	Command R+	77,554	4.3	1204
321	Qwen2-72B-Instruct	37,325	4.9	1203
322	Hunyuan Standard 256K	2,728	11.6	1202
323	Claude 3 Haiku (20240307)	117,701	3.7	1195
324	Llama 3.1 Tulu 3 8B	2,896	10.5	1193
325	DeepSeek-Coder-V2	15,147	6.3	1191
326	Ministral 8B (2410)	4,781	9	1191
327	Command R (08-2024)	10,140	6.5	1187
328	Jamba 1.5 Mini	8,858	7.2	1187
329	Llama 3.1 8B Instruct	49,605	4.1	1187
330	GPT-4 (0613)	88,723	4	1186
331	Aya Expanse 8B	9,818	6.9	1185
332	Mistral Large (2402)	62,436	4.7	1176
333	Qwen1.5-110B-Chat	26,195	5.5	1175
334	Yi 1.5 34B Chat	24,146	5	1173
335	Reka Flash 21B (Online, 20240226)	15,450	7.3	1170
336	Qwen1.5-72B-Chat	39,302	5.2	1166
337	Llama 3 8B Instruct	104,642	3.7	1166
338	Mistral Medium	34,550	5.4	1165
339	Reka Flash 21B (20240226)	24,806	5.9	1165
340	Command R	54,036	4.7	1163
341	Mixtral 8x22B Instruct v0.1	51,416	4.5	1162
342	QwQ-32B-Preview	3,231	11.3	1162
343	InternLM2.5 20B Chat	9,901	7.1	1158
344	Gemma 2 2B IT	46,616	4	1156
345	Granite 3.1 8B Instruct	3,090	10.9	1149
346	Gemini Pro (Dev API)	18,354	7.2	1149
347	Zephyr ORPO 141B A35B v0.1	4,652	10.7	1144
348	Phi-3 Medium 4K Instruct	25,055	5.1	1138
349	Qwen1.5-32B-Chat	21,741	6.1	1137
350	Starling LM 7B Beta	16,056	7.3	1132
351	Mixtral 8x7B Instruct v0.1	73,503	4.2	1132
352	Gemini Pro	6,390	11.4	1131
353	Yi 34B Chat	15,483	6.8	1128
354	Qwen1.5-14B-Chat	17,839	7	1128
355	Granite 3.1 2B Instruct	3,188	11	1127
356	GPT-3.5 Turbo (0125)	66,207	4.7	1125
357	Tulu 2 DPO 70B	6,535	9.7	1121
358	WizardLM 70B	8,214	9.3	1119
359	DBRX Instruct Preview	32,191	6	1119
360	Llama 2 70B Chat	38,492	5.4	1115
361	Nous Hermes 2 Mixtral 8x7B DPO	3,777	11.8	1112
362	Phi-3 Small 8K Instruct	17,766	5.8	1110
363	Llama 3.2 3B Instruct	7,936	7.6	1110
364	Starling LM 7B Alpha	10,224	7.9	1108
365	OpenChat 3.5 (0106)	12,637	7.9	1107
366	Vicuna 33B	22,479	6.1	1105
367	DeepSeek LLM 67B Chat	4,932	11.5	1105
368	Snowflake Arctic Instruct	32,832	5.9	1101
369	Llama 2 70B SteerLM Chat	3,585	12.4	1098
370	OpenChat 3.5	7,968	9.6	1097
371	Granite 3.0 8B Instruct	6,638	8.5	1096
372	GPT-3.5 Turbo (1106)	16,619	8.7	1094
373	Gemma 1.1 7B IT	23,893	6	1094
374	OpenHermes 2.5 Mistral 7B	5,006	10.3	1092
375	Mistral 7B Instruct v0.2	19,402	6.5	1090
376	Llama 2 13B Chat	19,174	6.7	1084
377	Qwen1.5-7B-Chat	4,737	9.7	1083
378	SOLAR 10.7B Instruct v1.0	4,155	13	1083
379	Dolphin 2.2.1 Mistral 7B	1,679	15.3	1081
380	Phi-3 Mini 4K Instruct (June 2024)	12,297	6.3	1080
381	Granite 3.0 2B Instruct	6,837	8.2	1080
382	WizardLM 13B	7,044	9.2	1077
383	Phi-3 Mini 4K Instruct	20,118	6.3	1073
384	Zephyr 7B Beta	11,118	8.7	1069
385	MPT 30B Chat	2,572	12.1	1069
386	Code Llama 34B Instruct	7,366	8.8	1065
387	Zephyr 7B Alpha	1,785	15.7	1058
388	Vicuna 13B	19,367	6.6	1058
389	Code Llama 70B Instruct	1,143	18	1057
390	Gemma 7B IT	8,925	9.5	1056
391	Llama 3.2 1B Instruct	8,045	7.7	1055
392	Falcon 180B Chat	1,295	17	1054
393	Llama 2 7B Chat	14,148	7	1053
394	Guanaco 33B	2,921	12	1053
395	Qwen-14B-Chat	4,964	10.9	1051
396	Phi-3 Mini 128K Instruct	20,685	7.2	1050
397	SmolLM2 1.7B Instruct	2,199	14.1	1042
398	StripedHyena Nous 7B	5,182	10.9	1038
399	OLMo 7B Instruct	6,328	11	1032
400	Vicuna 7B	6,923	9.1	1031
401	PaLM 2	8,554	9.2	1027
402	Mistral 7B Instruct	8,977	9.2	1023
403	Gemma 1.1 2B IT	10,854	7.7	1022
404	Gemma 2B IT	4,780	11.5	1002
405	Qwen1.5-4B-Chat	7,597	9.2	997
406	Koala 13B	6,965	9.7	990
407	ChatGLM3 6B	4,658	11.6	972
408	GPT4All 13B Snoozy	1,743	15	956
409	MPT 7B Chat	3,924	11.7	955
410	RWKV-4 Raven 14B	4,845	11.2	948
411	ChatGLM2 6B	2,658	13.4	939
412	Alpaca 13B	5,745	11.3	933
413	ChatGLM 6B	4,914	12.2	918
414	OpenAssistant Pythia 12B	6,310	10.7	916
415	FastChat-T5 3B	4,203	12.2	894
416	StableLM Tuned Alpha 7B	3,287	12.5	867
417	Dolly 2.0 12B	3,412	13.2	851
418	LLaMA 13B	2,391	15.5	833

只取 overall（综合）维度，不含 coding / math / 多语言等子维度。 同一模型在多个发布日重复上榜时只保留最新一次。 区间（±）越宽说明票越少，名次越不可靠。 榜单发布日：2026-10-22