M
ModelSage
排行榜/Arena 人类盲测/Text 文本

Arena Text 文本排行榜

文本对话综合 Elo,真实用户匿名两两对战投票 · 完整榜单 415 个模型

RankModelElo 分
1
gemini-4-argon-high1524.8±8.8
2
muse-wonder-max1512.7±13.3
3
claude-opus-5.5-high1508.1±7.5
4
claude-fable-5-high1503.4±4.3
5
claude-opus-4-6-high1503.1±3.4
6
claude-opus-4-7-high1501±3.8
7
claude-fable-5.1-max1500.6±6.1
8
claude-opus-4-61497.7±3.3
9
gemini-3.8-flash-high1496.3±4.6
10
muse-spark-1.2 (xHigh)1494.6±7.4
11
muse-spark-1.3-max1494.2±5.7
12
claude-opus-4-71493.6±3.8
13
muse-spark-1.11490.2±4.3
14
claude-opus-5-high1489.8±3.6
15
claude-opus-5-max1489.5±4.9
16
muse-spark1489.1±5.8
17
kimi-k3-max1488.1±4.6
18
gemini-3.7-flash-high1487.4±4.9
19
gemini-3.1-pro-preview1487.1±3
20
gemini-3-pro1485.4±3.8
21
gpt-5.6-sol-xhigh1484.9±4.4
22
gpt-6.1-sol-max1484±8.4
23
gemini-3.6-flash-high1482.9±4.4
24
qwen3.8-max1482.2±5
25
claude-opus-4-8-high1481.5±3.9
26
gpt-5.5-high1481.4±3.8
27
mimo-v2.6-pro1478.8±9.4
28
glm-5.3-max1478.5±5.3
29
gemini-3.5-flash-high1477.3±3.9
30
gpt-6-astra-max1476.7±6.4
31
gpt-5.51476.7±3.7
32
gemini-3.5-flash-medium1475.8±4.1
33
gpt-5.2-chat-latest-202602101475.6±4
34
glm-5.2-max1475.5±4.2
35
claude-sonnet-5.5-xhigh1475.3±8
36
gpt-5.4-high1475.1±3.8
37
grok-4.20-beta11474.8±4.5
38
deepseek-v4.1-flash-max1474.6±6.3
39
qwen3.7-max-preview1474.4±9.7
40
claude-opus-4-81474.3±3.8
41
glm-5.3-flash1474.2±4.9
42
claude-opus-4-5-20251101-high-32k1473.7±3.9
43
gpt-5.5-instant1473.1±5
44
gemini-3-flash1472.9±4.4
45
claude-sonnet-4-61472.1±3.5
46
grok-4.20-beta-0309-reasoning1471.5±3.6
47
grok-4.20-multi-agent-beta-03091470.5±3.7
48
claude-opus-4-5-202511011469.7±3.1
49
mimo-v2.5-pro1467.4±3.5
50
ernie-5.11467.2±4.6
51
gpt-5.6-terra-xhigh1465.9±4.3
52
grok-4.51465.7±4.3
53
glm-5.11465.4±3.6
54
grok-4.1-thinking1465±3.1
55
gpt-5.41464.7±3.7
56
qwen3.5-max-preview1464.4±4.9
57
deepseek-v4-pro-high-202608131464.4±6.6
58
claude-sonnet-5-high1461.3±4.2
59
kimi-k2.61460.8±4.5
60
qwen3.6-max-preview1460.3±8.2
61
grok-4.11459.1±3.1
62
gemini-3-flash (thinking-minimal)1458.3±3.1
63
gpt-6-sol-max1458.3±6.4
64
deepseek-v4-pro1457.8±3.9
65
glm-51457.4±4.3
66
claude-sonnet-4-5-20250929-high-32k1456.4±2.8
67
dola-seed-2.0-pro1456.2±3.4
68
qwen3.7-plus1455.7±4.3
69
gpt-5.1-high1455.7±3.7
70
grok-4.6-high1455±4.9
71
claude-sonnet-4-5-202509291455±2.9
72
gemini-3.5-flash-lite1455±4.5
73
deepseek-v4-pro-high-preview1454.8±3.9
74
hy31452.7±6.2
75
gpt-5.6-luna-xhigh1452.5±4.3
76
gemma-4-31b1452.1±7.4
77
mimo-v2.6-flash1450.9±6.6
78
gpt-5.3-chat-latest1450.1±4.3
79
kimi-k2.5-thinking1449.9±3.3
80
claude-opus-4-1-20250805-thinking-16k1449.8±3.5
81
ernie-5.0-preview-12031448.7±6.5
82
claude-opus-4-1-202508051448±3
83
mimo-v2-pro1447.7±4.7
84
gpt-5.4-mini-high1447.1±3.8
85
ernie-5.0-01101446.3±3.9
86
Step 5 Preview1445.8±7.9
87
gemini-2.5-pro1445.4±2.5
88
gpt-4.5-preview-2025-02-271444.6±5.7
89
grok-4.7-xhigh1443.8±6.6
90
qwen3.6-plus1443.1±4.1
91
gpt-6-luna-max1443±4.6
92
chatgpt-4o-latest-202503261442.6±2.8
93
qwen3.5-397b-a17b1441.8±3.1
94
grok-4.31441.5±3.6
95
glm-4.71441.4±6.1
96
minimax-m31441.4±3.9
97
inkling1440.8±4.3
98
deepseek-v4-flash-high-preview1438.7±4.1
99
gpt-5.11438.5±3.5
100
qwen3.8-27b1437.7±5.1
101
gemma-4-26b-a4b1437.3±7.4
102
gpt-5.2-high1437.2±3.5
103
longcat-flash-chat-2602-exp1436.3±4.5
104
deepseek-v4-flash1436±4
105
gpt-5.21435.4±3.1
106
gpt-5-high1434.8±4.5
107
qwen3-max-preview1434.2±4.5
108
mimo-v2.51433.8±4.2
109
glm-5v-turbo1433.7±6.4
110
gemini-3.1-flash-lite-preview1432.9±3.6
111
o3-2025-04-161431.7±3.6
112
mistral-large-41431.5±7.6
113
mimo-v2-omni1430.7±5.6
114
kimi-k2.5-instant1430.6±6.5
115
grok-4-1-fast-reasoning1430.5±3.2
116
kimi-k2-thinking-turbo1430.1±3.1
117
mistral-medium-3.51426.6±6.4
118
amazon-nova-experimental-chat-26-02-101426.4±9.7
119
gpt-5-chat1426.3±4.3
120
claude-opus-4-20250514-thinking-16k1426.2±4.4
121
nvidia-nemotron-3-ultra-550b-a55b-nvfp41425.7±6.9
122
deepseek-v3.2-exp-thinking1424.9±6.6
123
deepseek-v3.21424.7±3.5
124
muse-glimmer1424.5±9.7
125
glm-4.61424.1±3.9
126
qwen3-max-2025-09-231423.3±6.5
127
deepseek-v3.2-thinking1422.5±3.5
128
qwen3-235b-a22b-instruct-25071422.1±2.6
129
deepseek-v3.2-exp1421.6±6.4
130
deepseek-r1-05281421.4±5.7
131
kimi-k2-0905-preview1418.7±6.6
132
grok-4-fast-chat1418.5±7.8
133
ernie-5.0-preview-10221418.3±8.8
134
kimi-k2-0711-preview1417.9±4.9
135
deepseek-v3.11417.1±6.1
136
qwen3.5-122b-a10b1416.4±4.3
137
deepseek-v3.1-terminus-thinking1416.3±10.1
138
deepseek-v3.1-thinking1415.8±6.6
139
deepseek-v3.1-terminus1414.9±9.7
140
minimax-m2.71414.6±3.4
141
gpt-4.1-2025-04-141414.5±3.7
142
claude-opus-4-202505141414.2±4.4
143
mistral-large-31414.1±2.9
144
claude-haiku-4-5-202510011414.1±2.4
145
qwen3-vl-235b-a22b-instruct1413.4±6.5
146
amazon-nova-experimental-chat-26-01-101412.8±9.7
147
grok-3-preview-02-241411.2±4.4
148
grok-4-07091410.9±3.9
149
glm-4.51410.8±4.9
150
gemini-2.5-flash1409.3±2.5
151
hunyuan-hy3-preview1409.1±7.4
152
qwen3.5-27b1408.8±4.3
153
mistral-medium-25081407.8±2.6
154
grok-4-fast-reasoning1405.3±5
155
Inkling Small1403.5±4.7
156
gemini-2.5-flash-preview-09-20251403±4
157
o1-2024-12-171402.3±4.5
158
qwen3-235b-a22b-no-thinking1402.1±4.6
159
claude-sonnet-4-20250514-thinking-32k1401.9±4.4
160
gpt-5.4-nano-high1401.3±3.8
161
longcat-flash-chat1400.8±6.4
162
qwen3-235b-a22b-thinking-25071399.7±6.6
163
qwen3-next-80b-a3b-instruct1398.5±4.8
164
deepseek-r11398.1±4.9
165
qwen3.5-flash1396.1±3.6
166
deepseek-v3-03241395.2±3.9
167
qwen3-vl-235b-a22b-thinking1394.8±6.9
168
hunyuan-vision-1.5-thinking1393.9±12.4
169
qwen3.5-35b-a3b1393.8±4.2
170
step-3.5-flash1393±3.5
171
amazon-nova-experimental-chat-12-101392.9±9.4
172
mimo-v2-flash (non-thinking)1391.6±3.5
173
o4-mini-2025-04-161390.9±4
174
claude-sonnet-4-202505141390.9±4.4
175
minimax-m2.51390.7±3.9
176
gpt-5-mini-high1389.9±4.6
177
o1-preview1388.6±5
178
claude-3-7-sonnet-20250219-thinking-32k1388.3±4.2
179
qwen3-coder-480b-a35b-instruct1387.6±5
180
mistral-medium-25051387.2±4.8
181
hunyuan-t1-202507111387±8.7
182
solar-pro41387±5.4
183
mimo-v2-flash (thinking)1386.3±6.2
184
minimax-m2.1-preview1384.5±5.2
185
hunyuan-turbos-202504161383.1±6.5
186
gpt-4.1-mini-2025-04-141382.8±4.3
187
qwen3-30b-a3b-instruct-25071382.2±4.9
188
gemini-2.5-flash-lite-preview-09-2025-no-thinking1379.3±3.4
189
trinity-large-preview1378.7±4.3
190
glm-4.6v1378.2±11.2
191
qwen3-235b-a22b1374.8±4.7
192
gemini-2.5-flash-lite-preview-06-17-thinking1374.7±4.6
193
claude-3-5-sonnet-202410221374.3±3.2
194
qwen2.5-max1373.7±4.1
195
glm-4.5-air1373±4.3
196
claude-3-7-sonnet-202502191372.5±4
197
qwen3-next-80b-a3b-thinking1369.3±5.8
198
trinity-large-thinking1367.3±4.6
199
gemma-3-27b-it1365.4±3.7
200
glm-4.7-flash1364.6±5.8
201
amazon-nova-experimental-chat-11-101363.9±4.3
202
minimax-m11363.8±4.3
203
grok-3-mini-high1363.8±5.5
204
o3-mini-high1363.6±5.3
205
nvidia-nemotron-3-super-120b-a12b1360.1±7.1
206
gemini-2.0-flash-0011359.9±3.8
207
deepseek-v31358.4±4.8
208
grok-3-mini-beta1357.8±5.2
209
mistral-small-25061356.1±5.2
210
intellect-31355.7±8.3
211
command-a-03-20251353.7±3.5
212
gemini-2.0-flash-lite-preview-02-051353.5±4.4
213
glm-4.5v1352.1±8.4
214
gpt-oss-120b1351.7±4.4
215
gemini-1.5-pro-0021351.3±3.4
216
nvidia-nemotron-3.5-lightning-30b-a3b-nvfp41351.2±5.6
217
step-31348.9±7.6
218
hunyuan-turbos-202502261348.8±11.7
219
o3-mini1348±3.5
220
amazon-nova-experimental-chat-10-201347.9±6.1
221
llama-3.1-nemotron-ultra-253b-v11347.6±11.6
222
qwen3-32b1346.8±9.5
223
amazon-nova-experimental-chat-10-091346.1±10.9
224
gpt-4o-2024-05-131346.1±3.4
225
qwen-plus-01251345.9±8.4
226
minimax-m21343.4±7.8
227
ling-flash-2.01343.4±7.3
228
claude-3-5-sonnet-202406201343.3±3.5
229
mercury-21343.2±10.5
230
nvidia-llama-3.3-nemotron-super-49b-v1.51343±10
231
glm-4-plus-01111342.7±8.4
232
gemma-3-12b-it1341.7±9.5
233
hunyuan-turbo-01101340.6±11.6
234
granite-4.2-30b1339.8±10.5
235
gpt-5-nano-high1337.7±6.9
236
o1-mini1336.8±3.7
237
gemini-advanced-05141336±5.2
238
grok-2-2024-08-131335.6±3.6
239
gpt-4o-2024-08-061335.6±4.2
240
qwq-32b1335.5±4.5
241
llama-3.1-405b-instruct-bf161335.5±3.8
242
nova-2-lite1334.8±6.1
243
llama-3.1-405b-instruct-fp81333.6±3.6
244
step-2-16k-exp-2024121333.5±8.6
245
olmo-3.1-32b-instruct1329.1±6.1
246
yi-lightning1328.3±5
247
llama-3.3-nemotron-49b-super-v11327.7±12.1
248
llama-4-maverick-17b-128e-instruct1326.9±4.3
249
qwen3-30b-a3b1326.6±4.8
250
hunyuan-large-2025-02-101326.3±9.7
251
claude-3-5-haiku-202410221324.8±3.3
252
gpt-4-turbo-2024-04-091324.5±3.9
253
gemini-1.5-pro-0011324.1±4
254
deepseek-v2.5-12101323.4±8.3
255
ring-flash-2.01322.7±7.3
256
gpt-4.1-nano-2025-04-141322.3±7.8
257
claude-3-opus-202402291322.2±3
258
molmo-2-8b1321.9±21.1
259
llama-4-scout-17b-16e-instruct1321.5±4.8
260
step-1o-turbo-2025061319.7±6.9
261
glm-4-plus1319.3±4.9
262
qwen-max-09191318.1±5.7
263
llama-3.3-70b-instruct1318.1±3.5
264
gemma-3n-e4b-it1317.6±5.2
265
gpt-4o-mini-2024-07-181317.6±3.5
266
gpt-oss-20b1317.4±6.4
267
qwen2.5-plus-11271314.6±6.4
268
mistral-large-24071314.4±4
269
athene-v2-chat1314.1±4.6
270
nvidia-nemotron-3-nano-30b-a3b-bf161313.6±5.5
271
gpt-4-0125-preview1313.3±4.1
272
gpt-4-1106-preview1312.9±3.9
273
hunyuan-standard-2025-02-101311.3±9.7
274
gemini-1.5-flash-0021309.3±4.2
275
grok-2-mini-2024-08-131308.4±3.8
276
deepseek-v2.51307.2±4.7
277
athene-70b-07251306.7±5.6
278
olmo-3-32b-think1306.1±8.3
279
mercury1305.9±13.8
280
mistral-large-24111305.6±4.5
281
magistral-medium-25061304.7±6.4
282
granite-4.1-8b1303.8±10
283
gemma-3-4b-it1303.2±9.3
284
qwen2.5-72b-instruct1302.8±4.1
285
mistral-small-3.1-24b-instruct-25031302.7±4.6
286
llama-3.1-nemotron-70b-instruct1298.7±7.8
287
hunyuan-large-vision1293.7±9.2
288
llama-3.1-70b-instruct1293.4±3.8
289
granite-4.2-8b1290.2±11.1
290
amazon-nova-pro-v1.01290.1±4.6
291
gemma-2-27b-it1289.5±3.4
292
jamba-1.5-large1289.5±7.3
293
granite-4.2-3b1289.4±11.6
294
reka-core-202409041288±7.1
295
gpt-4-03141287.9±4.8
296
ibm-granite-h-small1287.4±8.4
297
llama-3.1-nemotron-51b-instruct1286.7±10
298
olmo-3.1-32b-think1286.6±7.3
299
gemini-1.5-flash-0011286.6±4.5
300
llama-3.1-tulu-3-70b1286.1±10.4
301
claude-3-sonnet-202402291281.4±4
302
gemma-2-9b-it-simpo1280.4±6.9
303
nemotron-4-340b-instruct1277.2±5.3
304
llama-3-70b-instruct1276.6±3.6
305
command-r-plus-08-20241276.4±6.6
306
gpt-4-06131276.3±4.1
307
mistral-small-24b-instruct-25011274.4±5.9
308
glm-4-05201273.5±7
309
reka-flash-202409041272.5±7
310
qwen2.5-coder-32b-instruct1270.6±8.1
311
c4ai-aya-expanse-32b1267.3±4.9
312
gemma-2-9b-it1267±3.9
313
deepseek-coder-v21265.4±6.3
314
qwen2-72b-instruct1261.9±4.9
315
command-r-plus1261.8±4.4
316
claude-3-haiku-202403071261.7±3.8
317
amazon-nova-lite-v1.01260.4±5.2
318
gemini-1.5-flash-8b-0011258.8±4.3
319
phi-41256.3±4.6
320
olmo-2-0325-32b-instruct1251.7±10.8
321
command-r-08-20241250.5±6.6
322
mistral-large-24021242.6±4.7
323
amazon-nova-micro-v1.01240.7±5.2
324
jamba-1.5-mini1239.8±7.2
325
ministral-8b-24101237.6±9.1
326
gemini-pro-dev-api1236.9±7.3
327
qwen1.5-110b-chat1234.2±5.5
328
reka-flash-21b-20240226-online1233.6±7.4
329
qwen1.5-72b-chat1233.6±5.3
330
hunyuan-standard-256k1233.2±11.7
331
mixtral-8x22b-instruct-v0.11229.7±4.6
332
command-r1227±4.8
333
reka-flash-21b-202402261226.9±5.9
334
gpt-3.5-turbo-01251225.7±4.7
335
gemini-pro1223.7±11.7
336
llama-3-8b-instruct1223.7±3.7
337
c4ai-aya-expanse-8b1223.1±7
338
mistral-medium1222.8±5.5
339
llama-3.1-tulu-3-8b1220.5±10.7
340
zephyr-orpo-141b-A35b-v0.11213.1±10.8
341
yi-1.5-34b-chat1212.9±5
342
llama-3.1-8b-instruct1211.2±4.2
343
granite-3.1-8b-instruct1208.6±11.1
344
gpt-3.5-turbo-11061204.4±8.8
345
qwen1.5-32b-chat1204.1±6.1
346
gemma-2-2b-it1200.1±4.1
347
phi-3-medium-4k-instruct1198±5.1
348
mixtral-8x7b-instruct-v0.11197.3±4.3
349
dbrx-instruct-preview1195.6±6.1
350
qwen1.5-14b-chat1191.2±7.1
351
internlm2_5-20b-chat1190.8±7.2
352
deepseek-llm-67b-chat1185±11.5
353
wizardlm-70b1184.8±9.4
354
yi-34b-chat1183.8±6.8
355
granite-3.0-8b-instruct1183.2±8.7
356
openchat-3.51183.2±9.6
357
openchat-3.5-01061182.8±7.9
358
gemma-1.1-7b-it1182.7±6.1
359
snowflake-arctic-instruct1180.2±5.9
360
granite-3.1-2b-instruct1178.9±11.1
361
tulu-2-dpo-70b1177.7±9.8
362
openhermes-2.5-mistral-7b1176.1±10.4
363
vicuna-33b1173±6.2
364
phi-3-small-8k-instruct1171.2±5.9
365
starling-lm-7b-beta1171±7.4
366
llama-2-70b-chat1170.7±5.5
367
starling-lm-7b-alpha1167.5±8
368
llama-3.2-3b-instruct1166.8±7.7
369
nous-hermes-2-mixtral-8x7b-dpo1164.5±11.9
370
granite-3.0-2b-instruct1156.8±8.3
371
llama2-70b-steerlm-chat1154.8±12.6
372
qwq-32b-preview1154.2±11.5
373
solar-10.7b-instruct-v1.01152.5±13.1
374
dolphin-2.2.1-mistral-7b1152.4±15.3
375
mpt-30b-chat1150.9±12.2
376
wizardlm-13b1149.6±9.2
377
mistral-7b-instruct-v0.21149.5±6.6
378
falcon-180b-chat1148.2±17
379
qwen1.5-7b-chat1143.9±9.9
380
phi-3-mini-4k-instruct-june-20241143.2±6.4
381
vicuna-13b1141.7±6.6
382
llama-2-13b-chat1141.5±6.7
383
qwen-14b-chat1139.5±11
384
palm-21139.3±9.3
385
gemma-7b-it1137.8±9.5
386
codellama-34b-instruct1137±8.9
387
zephyr-7b-beta1130.9±8.8
388
phi-3-mini-128k-instruct1129.9±7.4
389
phi-3-mini-4k-instruct1128.2±6.4
390
guanaco-33b1127.6±12
391
zephyr-7b-alpha1127±15.5
392
stripedhyena-nous-7b1121.7±11
393
codellama-70b-instruct1119.2±18
394
gemma-1.1-2b-it1116.8±7.8
395
vicuna-7b1115.4±9.1
396
smollm2-1.7b-instruct1114.8±14.3
397
llama-3.2-1b-instruct1111.2±7.9
398
mistral-7b-instruct1110.5±9.3
399
llama-2-7b-chat1108±7
400
gemma-2b-it1093.7±11.5
401
qwen1.5-4b-chat1091.2±9.3
402
olmo-7b-instruct1073.7±11.1
403
koala-13b1071±9.8
404
alpaca-13b1070.4±11.4
405
gpt4all-13b-snoozy1067.8±15.1
406
mpt-7b-chat1063.6±11.9
407
chatglm3-6b1056.7±11.6
408
RWKV-4-Raven-14B1042.4±11.4
409
chatglm2-6b1024.9±13.6
410
oasst-pythia-12b1023.6±10.8
411
chatglm-6b995.9±12.5
412
fastchat-t5-3b992.7±12.3
413
dolly-v2-12b982.6±13.4
414
llama-13b975.3±15.7
415
stablelm-tuned-alpha-7b953.3±12.6
Elo 分 真人在不知道模型名字的情况下两两投票,胜负累积出的等级分。绝对值没有意义,差值才有:分差 100 分 ≈ 强的那个有 64% 胜率,只差几分基本是平手。
± 后面的数 是 95% 置信区间的半宽,投票越多越小。两个模型的区间重叠,说明它们的名次差别在统计上不成立。
区间 把上面这个置信区间换算成名次:1–5 意思是这个模型的真实名次可能在第 1 到第 5 之间。
数据更新: 2026-10-14数据来源: Arena AI ↗