Benchmarks / Arena (formerly LMArena)

Reported by Arena (formerly LMArena)

Arena (formerly LMArena)

Human preference on writing and language prompts.

Last updated 2 Oct 2026

Results dated
2 Oct 2026
Results
412 configurations of 386 models
Unit
Arena rating
Licence
Creative Commons Attribution 4.0 International

Arena (formerly LMArena) in brief

What it measures. Arena (formerly LMArena) is a third-party benchmark, shown here with credit. Measured here: Human preference on writing and language prompts.

Latest results. , 412 configurations ranked.

Top 3 on Writing, literature and language (Arena rating, higher is better):

  1. Gemini 4 Argon (high reasoning) (Google): 1,522
  2. Claude Opus 5.5 (high reasoning) (Anthropic): 1,519
  3. Claude Fable 5 (high reasoning) (Anthropic): 1,508

The first 4 are too close to tell apart, so read the order among them as a tie.

How to cite. Arena (formerly LMArena), via Spring Prompt, results dated 2 October 2026. https://springprompt.com/benchmarks/lmarena These results belong to Arena (formerly LMArena), not to us. Licence: Creative Commons Attribution 4.0 International. Credit Arena (formerly LMArena) first.

Writing, literature and language:

Top 15 of 412 results · Arena rating, higher is better · lines show the 95% range · ≈ cannot be told apart from the leader. Choose a model to highlight it.Clear highlight

  1. 1≈ Gemini 4 Argon (high reasoning)Google 1,522
  2. 2≈ Claude Opus 5.5 (high reasoning)Anthropic 1,519
  3. 3≈ Claude Fable 5 (high reasoning)Anthropic 1,508
  4. 4≈ Claude Opus 4.6 (high reasoning)Anthropic 1,500
  5. 5 Claude Opus 4.7 (high reasoning)Anthropic 1,497
  6. 6 Gemini 3.7 Flash (high reasoning)Google 1,496
  7. 7 Claude Fable 5.1 (max reasoning)Anthropic 1,491
  8. 8 Claude Opus 4.6Anthropic 1,490
  9. 9 Gemini 3.8 Flash (high reasoning)Google 1,487
  10. 10 Claude Opus 4.7Anthropic 1,484
  11. 11 GPT-5.6 Sol (extra-high reasoning)OpenAI 1,481
  12. 11 Gemini 3 Pro PreviewGoogle 1,481
  13. 13 Gemini 3.1 Pro PreviewGoogle 1,480
  14. 13 Claude Opus 5 (high reasoning)Anthropic 1,480
  15. 13 Claude Opus 5 (max reasoning)Anthropic 1,480

Full results

Arena Text: writing, literature and language, Arena rating, higher is better
#ModelWriting, literature and language · 95% range
Arena rating, higher is better
Price
$ per million tokens, in / out
1≈ Gemini 4 Argon (high reasoning)Google
1,522
1,505–1,539
$2 / $10
2≈ Claude Opus 5.5 (high reasoning)Anthropic
1,519
1,502–1,536
$4 / $20
3≈ Claude Fable 5 (high reasoning)Anthropic
1,508
1,501–1,514
$10 / $50
4≈ Claude Opus 4.6 (high reasoning)Anthropic
1,500
1,494–1,505
$5 / $25
5 Claude Opus 4.7 (high reasoning)Anthropic
1,497
1,491–1,503
$5 / $25
6 Gemini 3.7 Flash (high reasoning)Google
1,496
1,487–1,504
$1.50 / $7.50
7 Claude Fable 5.1 (max reasoning)Anthropic
1,491
1,480–1,502
$10 / $50
8 Claude Opus 4.6Anthropic
1,490
1,485–1,495
$5 / $25
9 Gemini 3.8 Flash (high reasoning)Google
1,487
1,479–1,495
$1.50 / $7.50
10 Claude Opus 4.7Anthropic
1,484
1,478–1,490
$5 / $25
11 GPT-5.6 Sol (extra-high reasoning)OpenAI
1,481
1,474–1,489
$4 / $20
11 Gemini 3 Pro PreviewGoogle
1,481
1,474–1,488
–
13 Gemini 3.1 Pro PreviewGoogle
1,480
1,475–1,485
$2 / $12
13 Claude Opus 5 (high reasoning)Anthropic
1,480
1,474–1,486
$5 / $25
13 Claude Opus 5 (max reasoning)Anthropic
1,480
1,472–1,488
$5 / $25
16 Kimi K3 (max reasoning)Moonshot AI
1,476
1,468–1,484
$3 / $15
17 Claude Opus 4.8 (high reasoning)Anthropic
1,475
1,469–1,481
$5 / $25
18 Gemini 3.5 Flash (high reasoning)Google
1,474
1,467–1,480
$1.50 / $9
18 Gemini 3.6 Flash (high reasoning)Google
1,474
1,466–1,481
$1.50 / $7.50
20 GPT-6.1 Sol (max reasoning)OpenAI
1,472
1,452–1,493
$2 / $10
20 Qwen3.8-Max (0902)Alibaba
1,472
1,464–1,480
$2 / $6
20 Gemini 3.5 Flash (medium reasoning)Google
1,472
1,465–1,478
$1.50 / $9
23 Muse Spark 1.3 (max reasoning)Meta
1,470
1,460–1,481
$1.25 / $4.25
23 Muse Spark 1.2 (extra-high reasoning)Meta
1,470
1,452–1,487
$1.25 / $4.25
25 GPT-5.5 (high reasoning)OpenAI
1,469
1,463–1,475
$5 / $30
26 Claude Opus 4.8Anthropic
1,467
1,462–1,473
$5 / $25
26 Claude Opus 4.5Anthropic
1,467
1,461–1,472
$5 / $25
28 GPT-5.5OpenAI
1,466
1,461–1,472
$5 / $30
28 GLM-5.3 (max reasoning)Z.ai
1,466
1,457–1,475
$1.40 / $4.40
28 Claude Opus 4.5 (high reasoning, 32k budget)Anthropic
1,466
1,459–1,473
$5 / $25
31 GPT-6 Astra (max reasoning)OpenAI
1,465
1,453–1,478
$10 / $50
31 GPT-5.4 (high reasoning)OpenAI
1,465
1,459–1,471
$2.50 / $15
33 Muse SparkMeta
1,463
1,452–1,473
–
34 Claude Sonnet 5.5 (extra-high reasoning)Anthropic
1,461
1,441–1,482
$2 / $10
34 Gemini 3 Flash PreviewGoogle
1,461
1,453–1,469
$0.50 / $3
34 GPT-5.5 InstantOpenAI
1,461
1,453–1,469
–
34 Muse Spark 1.1Meta
1,461
1,454–1,468
$1.25 / $4.25
38 GLM-5.2 (max reasoning)Z.ai
1,459
1,453–1,466
$1.40 / $4.40
38 MiMo-V2.6-ProXiaomi
1,459
1,440–1,478
$0.43 / $0.87
38 Qwen3.7-Max-PreviewAlibaba
1,459
1,438–1,479
–
41 GPT-5.4OpenAI
1,458
1,452–1,464
$2.50 / $15
41 Grok 4.5xAI
1,458
1,451–1,465
$2 / $6
43 Qwen3.5-Max-PreviewAlibaba
1,455
1,447–1,464
–
43 Claude Sonnet 4.6Anthropic
1,455
1,449–1,461
$3 / $15
45 DeepSeek-V4.1-Flash (max reasoning)DeepSeek
1,454
1,442–1,467
$0.30 / $1.20
45 Claude Sonnet 4.5Anthropic
1,454
1,449–1,459
$3 / $15
45 DeepSeek-V4-Pro (0813, high reasoning)DeepSeek
1,454
1,442–1,466
$1.32 / $3.96
48 GPT-5.2 Chat (2026-02-10)OpenAI
1,453
1,446–1,460
–
49 Grok 4.20 Beta 1xAI
1,452
1,444–1,460
–
49 GLM-5.1Z.ai
1,452
1,446–1,458
$1.33 / $4.18
49 Claude Sonnet 4.5 (high reasoning, 32k budget)Anthropic
1,452
1,447–1,457
$3 / $15
52 Claude Sonnet 5 (high reasoning)Anthropic
1,451
1,445–1,458
$2 / $10
52 DeepSeek-V4-Pro (0423)DeepSeek
1,451
1,444–1,457
$1.42 / $2.83
54 MiMo-V2.5-ProXiaomi
1,450
1,445–1,456
$0.43 / $0.87
55 GLM-5Z.ai
1,449
1,441–1,456
$1 / $3.20
56 Gemini 3 Flash Preview (minimal reasoning)Google
1,448
1,443–1,453
$0.50 / $3
56 Grok 4.20 Beta (0309, reasoning)xAI
1,448
1,442–1,454
–
56 GLM-5.3-FlashZ.ai
1,448
1,440–1,456
$0.15 / $0.50
56 GPT-4.5 PreviewOpenAI
1,448
1,438–1,457
–
60 Qwen3.6-Max-PreviewAlibaba
1,446
1,429–1,464
–
60 Claude Opus 4.1 (16k reasoning budget)Anthropic
1,446
1,440–1,452
$15 / $75
62 Grok 4.6 (high reasoning)xAI
1,445
1,437–1,454
$2 / $6
62 Gemini 2.5 ProGoogle
1,445
1,440–1,449
$1.25 / $10
62 GPT-5.6 Terra (extra-high reasoning)OpenAI
1,445
1,438–1,451
$2 / $12
65 Claude Opus 4.1Anthropic
1,444
1,439–1,450
$15 / $75
66 Grok 4.20 Multi-Agent Beta (0309)xAI
1,443
1,437–1,449
–
66 DeepSeek-V4-Pro (0423, high reasoning)DeepSeek
1,443
1,437–1,449
$1.42 / $2.83
66 Gemini 3.5 Flash-LiteGoogle
1,443
1,436–1,450
$0.30 / $2.50
69 ERNIE 5.1Baidu
1,442
1,435–1,449
–
70 Kimi K2.6Moonshot AI
1,441
1,434–1,448
$0.95 / $4
71 GPT-6 Sol (max reasoning)OpenAI
1,439
1,426–1,451
$2 / $10
72 GPT-5.1 (high reasoning)OpenAI
1,438
1,431–1,444
$1.25 / $10
73 Qwen3.7-PlusAlibaba
1,437
1,430–1,444
$0.32 / $1.28
74 Step 5 PreviewStepFun
1,436
1,415–1,457
$1 / $2.70
75 Gemma 4 31BGoogle
1,435
1,420–1,451
$0.15 / $0.40
76 GPT-5.6 Luna (extra-high reasoning)OpenAI
1,433
1,426–1,440
$0.20 / $1.20
77 ChatGPT-4o (2025-03-26)OpenAI
1,431
1,426–1,436
–
77 Claude Opus 4 (16k reasoning budget)Anthropic
1,431
1,424–1,439
–
77 MiMo-V2-ProXiaomi
1,431
1,423–1,439
–
80 Grok 4.7 (extra-high reasoning)xAI
1,430
1,416–1,445
$2 / $6
81 GPT-5.3 ChatOpenAI
1,428
1,420–1,435
–
81 Kimi K2.5 (reasoning on)Moonshot AI
1,428
1,422–1,433
$0.57 / $2.85
83 Hy3Tencent
1,427
1,416–1,439
$0.14 / $0.58
83 Grok 4.1 ThinkingxAI
1,427
1,422–1,433
–
83 Grok 4.1xAI
1,427
1,421–1,433
–
86 GPT-5.1OpenAI
1,425
1,418–1,431
$1.25 / $10
87 Grok 4.3xAI
1,424
1,418–1,430
$1.25 / $2.50
88 GPT-5.4 mini (high reasoning)OpenAI
1,423
1,417–1,429
$0.75 / $4.50
88 Qwen3.6-PlusAlibaba
1,423
1,416–1,430
$0.33 / $1.95
90 MiniMax-M3MiniMax
1,422
1,416–1,428
$0.30 / $1.20
90 DeepSeek-V4-Flash (0423, high reasoning)DeepSeek
1,422
1,415–1,429
$0.05 / $0.45
92 GPT-6 Luna (max reasoning)OpenAI
1,421
1,409–1,433
$0.10 / $0.50
92 Qwen3.5-397B-A17BAlibaba
1,421
1,416–1,426
$0.55 / $3.50
94 Gemini 3.1 Flash-Lite PreviewGoogle
1,420
1,414–1,426
$0.25 / $1.50
94 Claude Opus 4Anthropic
1,420
1,413–1,426
–
96 MiMo-V2.6-FlashXiaomi
1,419
1,404–1,435
$0.14 / $0.28
96 ERNIE 5.0 (0110)Baidu
1,419
1,412–1,426
–
98 GLM-4.7Z.ai
1,416
1,405–1,427
$0.60 / $2.20
98 DeepSeek-V4-Flash (0423)DeepSeek
1,416
1,409–1,422
$0.05 / $0.45
100 ERNIE 5.0 Preview (1203)Baidu
1,415
1,403–1,428
–
101 GLM-5V-TurboZ.ai
1,414
1,402–1,426
$1.20 / $4
102 MiMo-V2.5Xiaomi
1,412
1,405–1,419
$0.40 / $2
102 DeepSeek-V3.1-Terminus (reasoning on)DeepSeek
1,412
1,390–1,433
$0.27 / $1
102 GPT-5.2OpenAI
1,412
1,406–1,417
$1.75 / $14
105 DeepSeek-V3.2-ExpDeepSeek
1,411
1,399–1,422
$0.27 / $0.41
106 DeepSeek-V3.2DeepSeek
1,410
1,403–1,416
$0.28 / $0.42
107 Dola-Seed-2.0-ProByteDance
1,409
1,404–1,415
–
108 GPT-5.2 (high reasoning)OpenAI
1,408
1,402–1,415
$1.75 / $14
108 DeepSeek-V3.1-TerminusDeepSeek
1,408
1,387–1,429
$0.27 / $1
108 Grok 4.1 Fast (reasoning)xAI
1,408
1,402–1,413
–
111 DeepSeek-V3.1 (reasoning on)DeepSeek
1,407
1,396–1,419
$0.27 / $1
112 GPT-5 ChatOpenAI
1,406
1,399–1,414
–
112 Grok 3 Preview (02-24)xAI
1,406
1,399–1,413
–
112 Gemma 4 26B A4BGoogle
1,406
1,391–1,421
$0.09 / $0.30
115 GPT-4.1OpenAI
1,405
1,399–1,412
$2 / $8
115 Qwen3-Max-PreviewAlibaba
1,405
1,398–1,413
–
115 Mistral Medium 3.5Mistral AI
1,405
1,394–1,417
$1.50 / $7.50
115 Qwen3-MaxAlibaba
1,405
1,392–1,418
–
115 GLM-4.6Z.ai
1,405
1,398–1,412
$0.50 / $2
115 Kimi K2.5 (no reasoning)Moonshot AI
1,405
1,391–1,418
$0.57 / $2.85
121 DeepSeek-V3.2 (reasoning on)DeepSeek
1,404
1,397–1,410
$0.28 / $0.42
122 ERNIE 5.0 Preview (1022)Baidu
1,402
1,385–1,419
–
122 Gemini 2.5 FlashGoogle
1,402
1,398–1,406
$0.30 / $2.50
122 MiMo-V2-OmniXiaomi
1,402
1,393–1,411
–
125 Claude Haiku 4.5Anthropic
1,400
1,396–1,404
$1 / $5
126 Kimi K2 Thinking TurboMoonshot AI
1,399
1,393–1,404
–
127 Grok 4 (0709)xAI
1,398
1,392–1,405
–
127 InklingThinking Machines
1,398
1,391–1,406
$1 / $4.05
127 o1OpenAI
1,398
1,391–1,406
–
127 Claude Sonnet 4 (32k reasoning budget)Anthropic
1,398
1,391–1,406
–
127 DeepSeek-V3.1DeepSeek
1,398
1,388–1,408
$0.27 / $1
127 DeepSeek-V3.2-Exp (reasoning on)DeepSeek
1,398
1,385–1,411
$0.27 / $0.41
133 Qwen3.8-27BAlibaba
1,397
1,388–1,407
$0.50 / $3
134 GPT-5 (high reasoning)OpenAI
1,396
1,389–1,404
$1.25 / $10
134 o3OpenAI
1,396
1,390–1,402
$2 / $8
134 Qwen3-235B-A22B-Instruct-2507Alibaba
1,396
1,392–1,401
$0.15 / $0.75
137 Gemini 2.5 Flash Preview (09-2025)Google
1,394
1,387–1,402
–
137 Claude 3.7 Sonnet (32k reasoning budget)Anthropic
1,394
1,387–1,401
–
137 Grok 4 Fast ChatxAI
1,394
1,378–1,409
–
140 LongCat-Flash-Chat (2602, experimental)Meituan
1,392
1,385–1,400
–
140 Nemotron 3 Ultra 550B A55B (NVFP4)NVIDIA
1,392
1,381–1,403
–
142 Claude Sonnet 4Anthropic
1,390
1,383–1,397
–
143 Mistral Large 3Mistral AI
1,389
1,384–1,394
$0.50 / $1.50
143 DeepSeek-R1-0528DeepSeek
1,389
1,379–1,398
$0.50 / $2.18
145 Grok 4 Fast (reasoning)xAI
1,388
1,379–1,397
–
146 Qwen3-VL-235B-A22B-InstructAlibaba
1,387
1,374–1,400
$0.21 / $1.90
146 Qwen3.5-122B-A10BAlibaba
1,387
1,380–1,395
$0.26 / $2.08
146 DeepSeek-V3-0324DeepSeek
1,387
1,380–1,393
$0.25 / $1
146 MiniMax-M2.7MiniMax
1,387
1,381–1,392
$0.30 / $1.20
150 Mistral Medium 3.1Mistral AI
1,386
1,382–1,391
$0.40 / $2
151 Muse GlimmerMeta
1,385
1,366–1,405
–
152 DeepSeek-R1DeepSeek
1,382
1,374–1,391
–
152 o1-previewOpenAI
1,382
1,374–1,390
–
152 GLM-4.5Z.ai
1,382
1,374–1,391
$0.60 / $2.20
152 Kimi K2 (0905)Moonshot AI
1,382
1,370–1,394
$0.60 / $2.50
156 Hunyuan Vision 1.5 ThinkingTencent
1,381
1,354–1,407
–
157 Qwen3.5-27BAlibaba
1,380
1,373–1,388
$0.27 / $2.16
158 Gemini 2.5 Flash-Lite Preview (06-17, reasoning on)Google
1,379
1,372–1,387
–
158 Claude 3.7 SonnetAnthropic
1,379
1,373–1,386
–
160 Qwen3-235B-A22B (no reasoning)Alibaba
1,378
1,370–1,385
–
160 Claude 3.5 Sonnet (2024-10-22)Anthropic
1,378
1,373–1,383
–
162 Qwen3-235B-A22B-Thinking-2507Alibaba
1,377
1,363–1,391
$0.45 / $3.50
162 Hy3 PreviewTencent
1,377
1,362–1,392
$0.18 / $0.60
164 Kimi K2 (0711)Moonshot AI
1,375
1,367–1,383
$0.57 / $2.30
165 Nova Experimental Chat (2026-02-10)Amazon
1,374
1,353–1,395
–
166 Gemini 2.5 Flash-Lite Preview (09-2025, no reasoning)Google
1,371
1,365–1,378
–
166 Nova Experimental Chat (12-10)Amazon
1,371
1,351–1,390
–
166 Mistral Medium 3Mistral AI
1,371
1,363–1,378
$0.40 / $2
169 MiniMax-M2.5MiniMax
1,369
1,363–1,376
$0.29 / $1.20
169 Nova Experimental Chat (2026-01-10)Amazon
1,369
1,348–1,389
–
169 Hunyuan T1 (2025-07-11)Tencent
1,369
1,350–1,387
–
172 Qwen3.5-FlashAlibaba
1,367
1,361–1,373
$0.065 / $0.26
173 Qwen3-Coder-480B-A35BAlibaba
1,366
1,358–1,375
$0.30 / $1
173 MiMo-V2-Flash (no reasoning)Xiaomi
1,366
1,360–1,372
–
173 Qwen3-VL-235B-A22B-ThinkingAlibaba
1,366
1,352–1,380
–
173 Step 3.5 FlashStepFun
1,366
1,360–1,372
$0.10 / $0.30
177 MiniMax-M2.1 PreviewMiniMax
1,364
1,355–1,374
–
178 Qwen3.5-35B-A3BAlibaba
1,363
1,356–1,371
$0.15 / $1
178 GPT-5.4 nano (high reasoning)OpenAI
1,363
1,357–1,369
$0.20 / $1.25
180 Qwen2.5-MaxAlibaba
1,362
1,356–1,369
–
181 GPT-4.1 miniOpenAI
1,361
1,354–1,368
$0.40 / $1.60
182 Trinity Large PreviewArcee AI
1,359
1,351–1,366
–
183 GLM-4.6VZ.ai
1,357
1,334–1,380
$0.30 / $0.90
184 Hunyuan TurboS (2025-04-16)Tencent
1,356
1,345–1,368
–
184 DeepSeek-V3DeepSeek
1,356
1,348–1,364
$0.26 / $1.03
184 MiMo-V2-Flash (reasoning on)Xiaomi
1,356
1,344–1,367
–
187 Gemma 3 27BGoogle
1,355
1,348–1,361
$0.08 / $0.35
188 Gemini 1.5 Pro (002)Google
1,354
1,348–1,360
–
188 LongCat-Flash-ChatMeituan
1,354
1,342–1,366
–
188 o4-miniOpenAI
1,354
1,348–1,361
–
188 Gemini 2.0 Flash (001)Google
1,354
1,347–1,360
–
192 GPT-5 mini (high reasoning)OpenAI
1,352
1,344–1,360
$0.25 / $2
193 Command ACohere
1,348
1,342–1,354
$2.50 / $10
193 Gemini 2.0 Flash-Lite Preview (02-05)Google
1,348
1,340–1,355
–
195 GPT-4o (2024-05-13)OpenAI
1,345
1,340–1,351
$5 / $15
195 Qwen3-235B-A22BAlibaba
1,345
1,337–1,353
–
197 Grok 3 Mini (high reasoning)xAI
1,344
1,334–1,354
–
198 Trinity Large ThinkingArcee AI
1,343
1,335–1,351
–
199 Qwen3-Next-80B-A3B-InstructAlibaba
1,342
1,334–1,351
$0.09 / $1.10
199 Inkling SmallThinking Machines
1,342
1,334–1,350
$0.45 / $1.20
199 Qwen3-30B-A3B-Instruct-2507Alibaba
1,342
1,333–1,350
$0.09 / $0.30
202 Qwen3-Next-80B-A3B-ThinkingAlibaba
1,341
1,331–1,352
–
202 o3-mini (high reasoning)OpenAI
1,341
1,333–1,350
–
202 GPT-4o (2024-08-06)OpenAI
1,341
1,334–1,348
$2.50 / $10
202 GLM-4.5-AirZ.ai
1,341
1,334–1,349
$0.14 / $0.86
202 Grok 3 Mini BetaxAI
1,341
1,332–1,350
–
202 Gemini Advanced (0514)Google
1,341
1,333–1,348
–
208 Solar Pro 4Upstage
1,340
1,330–1,350
$0.15 / $0.60
209 Gemini 1.5 Pro (001)Google
1,334
1,328–1,341
–
210 INTELLECT-3Prime Intellect
1,333
1,317–1,350
–
210 Hunyuan TurboS (2025-02-26)Tencent
1,333
1,312–1,354
–
210 MiniMax-M1MiniMax
1,333
1,325–1,340
–
213 GLM-4-Plus (0111)Z.ai
1,332
1,317–1,347
–
214 GLM-4.7-FlashZ.ai
1,331
1,320–1,343
$0.06 / $0.40
214 Claude 3.5 Sonnet (2024-06-20)Anthropic
1,331
1,325–1,337
–
216 GPT-4 TurboOpenAI
1,330
1,324–1,337
–
217 Mistral Small 3.2 24BMistral AI
1,329
1,319–1,339
$0.094 / $0.25
218 Nova Experimental Chat (11-10)Amazon
1,328
1,321–1,336
–
219 Grok 2 (2024-08-13)xAI
1,327
1,321–1,333
–
220 o3-miniOpenAI
1,325
1,319–1,331
–
221 Nemotron 3 Super 120B A12BNVIDIA
1,324
1,311–1,338
$0.08 / $0.45
222 Hunyuan Turbo (0110)Tencent
1,323
1,303–1,343
–
223 Qwen-PlusAlibaba
1,322
1,307–1,336
$0.26 / $0.78
224 Llama 3.1 Nemotron Ultra 253B v1NVIDIA
1,321
1,300–1,342
–
225 Step-2 16K Exp (2024-12)StepFun
1,319
1,304–1,335
–
226 Step 3StepFun
1,318
1,303–1,333
–
226 GLM-4.5VZ.ai
1,318
1,300–1,335
$0.60 / $1.80
228 Claude 3 OpusAnthropic
1,316
1,311–1,322
–
228 Llama 3.1 405B Instruct (FP8)Meta
1,316
1,310–1,322
–
230 Nova Experimental Chat (10-20)Amazon
1,315
1,304–1,327
–
230 DeepSeek-V2.5-1210DeepSeek
1,315
1,302–1,328
–
232 Claude 3.5 HaikuAnthropic
1,314
1,309–1,319
–
232 GPT-4o mini (2024-07-18)OpenAI
1,314
1,308–1,319
$0.15 / $0.60
234 Gemma 3 12BGoogle
1,313
1,296–1,331
$0.05 / $0.15
235 Llama 3.1 405B Instruct (BF16)Meta
1,312
1,306–1,318
–
235 Llama 4 MaverickMeta
1,312
1,305–1,319
$0.35 / $1
237 Llama 3.3 Nemotron Super 49B v1NVIDIA
1,311
1,288–1,334
–
237 o1-miniOpenAI
1,311
1,305–1,317
–
239 Hunyuan Large (2025-02-10)Tencent
1,310
1,291–1,329
–
240 GLM-4-PlusZ.ai
1,309
1,301–1,318
–
240 gpt-oss-120bOpenAI
1,309
1,302–1,317
$0.10 / $0.50
240 Gemini 1.5 Flash (002)Google
1,309
1,302–1,316
–
243 Qwen3-32BAlibaba
1,308
1,291–1,325
$0.14 / $0.40
243 Mercury 2Inception
1,308
1,286–1,330
–
243 QwQ-32BAlibaba
1,308
1,300–1,315
–
243 GPT-4 Turbo Preview (1106)OpenAI
1,308
1,301–1,314
–
247 Qwen-Max (0919)Alibaba
1,307
1,298–1,317
–
248 GPT-4 Turbo Preview (0125)OpenAI
1,306
1,299–1,312
–
249 Nova Experimental Chat (10-09)Amazon
1,305
1,282–1,329
–
249 Llama 4 ScoutMeta
1,305
1,297–1,313
$0.10 / $0.30
251 Qwen3-30B-A3BAlibaba
1,304
1,296–1,312
$0.12 / $0.50
252 Mistral Large 2 (2407)Mistral AI
1,303
1,296–1,310
$2 / $6
252 Yi-Lightning01.AI
1,303
1,295–1,311
–
254 Step-1o Turbo (2025-06)StepFun
1,302
1,288–1,315
–
255 Olmo 3.1 32B InstructAi2
1,300
1,288–1,312
–
255 Gemma 3n E4BGoogle
1,300
1,291–1,309
–
257 GPT-4.1 nanoOpenAI
1,299
1,285–1,314
$0.10 / $0.40
257 MiniMax-M2MiniMax
1,299
1,283–1,314
–
259 Mistral Large 2.1 (2411)Mistral AI
1,298
1,291–1,306
–
259 Nemotron 3.5 Lightning 30B A3B (NVFP4)NVIDIA
1,298
1,287–1,309
–
259 Nova 2 LiteAmazon
1,298
1,287–1,310
$0.30 / $2.50
259 Gemma 2 27BGoogle
1,298
1,292–1,303
$0.65 / $0.65
259 Gemma 3 4BGoogle
1,298
1,281–1,315
$0.04 / $0.05
259 Llama 3.3 Nemotron Super 49B v1.5NVIDIA
1,298
1,276–1,319
–
265 Grok 2 Mini (2024-08-13)xAI
1,295
1,288–1,301
–
266 Hunyuan Standard (2025-02-10)Tencent
1,293
1,275–1,312
–
267 Gemini 1.5 Flash (001)Google
1,292
1,285–1,299
–
267 Llama 3.3 70B InstructMeta
1,292
1,286–1,297
$0.71 / $0.71
269 Hunyuan Large VisionTencent
1,291
1,273–1,309
–
269 Magistral Medium 1.0Mistral AI
1,291
1,278–1,303
–
269 Mistral Small 3.1 24BMistral AI
1,291
1,283–1,298
$0.35 / $0.56
272 Ring-flash-2.0Ant Group
1,290
1,274–1,305
–
273 GPT-5 nano (high reasoning)OpenAI
1,289
1,275–1,303
$0.05 / $0.40
273 Qwen2.5-Plus (1127)Alibaba
1,289
1,278–1,300
–
275 Granite 4.2 30BIBM
1,287
1,266–1,309
–
276 GPT-4 (0613)OpenAI
1,285
1,278–1,292
–
277 DeepSeek-V2.5DeepSeek
1,284
1,276–1,292
–
277 GPT-4 (0314)OpenAI
1,284
1,275–1,292
–
279 Qwen2.5-72B-InstructAlibaba
1,282
1,276–1,289
$0.36 / $0.40
280 Athene 70B (0725)Nexusflow
1,279
1,270–1,288
–
280 Athene-V2-ChatNexusflow
1,279
1,271–1,286
–
282 Llama 3.1 Tülu 3 70BAi2
1,278
1,258–1,297
–
283 Ling-flash-2.0Ant Group
1,277
1,262–1,293
–
283 Jamba 1.5 LargeAI21 Labs
1,277
1,265–1,290
–
283 Command R+ (08-2024)Cohere
1,277
1,266–1,288
$2.50 / $10
286 Olmo 3 32B ThinkAi2
1,275
1,259–1,292
–
286 gpt-oss-20bOpenAI
1,275
1,262–1,288
$0.03 / $0.14
286 Gemma 2 9B SimPOPrinceton NLP
1,275
1,263–1,287
–
289 Nova Pro 1.0Amazon
1,274
1,267–1,281
$0.80 / $3.20
289 Granite 4.1 8BIBM
1,274
1,253–1,295
–
291 Reka Core (2024-09-04)Reka AI
1,272
1,258–1,286
–
292 Claude 3 SonnetAnthropic
1,270
1,263–1,276
–
292 Llama 3.1 Nemotron 51B InstructNVIDIA
1,270
1,251–1,288
–
294 Olmo 3.1 32B ThinkAi2
1,269
1,255–1,283
–
294 Gemma 2 9BGoogle
1,269
1,263–1,275
–
296 Nemotron 3 Nano 30B A3BNVIDIA
1,267
1,256–1,277
$0.06 / $0.24
297 Llama 3.1 Nemotron 70B InstructNVIDIA
1,265
1,251–1,279
–
298 Llama 3.1 70B InstructMeta
1,264
1,257–1,270
$0.72 / $0.72
299 Granite H SmallIBM
1,261
1,242–1,279
–
300 Command R+Cohere
1,259
1,252–1,267
–
300 Reka Flash (2024-09-04)Reka AI
1,259
1,246–1,272
–
302 Nemotron-4 340B InstructNVIDIA
1,258
1,249–1,268
–
303 MercuryInception
1,257
1,228–1,286
–
304 Gemini 1.5 Flash-8B (001)Google
1,255
1,248–1,262
–
305 GLM-4 (0520)Z.ai
1,254
1,242–1,266
–
305 Aya Expanse 32BCohere
1,254
1,247–1,262
–
307 Llama 3 70B InstructMeta
1,250
1,243–1,256
–
308 Mistral Small 3Mistral AI
1,248
1,239–1,258
$0.05 / $0.08
309 Claude 3 HaikuAnthropic
1,247
1,241–1,253
–
310 DeepSeek-Coder-V2DeepSeek
1,243
1,232–1,253
–
311 Qwen2-72B-InstructAlibaba
1,242
1,234–1,250
–
312 Nova Lite 1.0Amazon
1,240
1,231–1,248
$0.06 / $0.24
313 Command R (08-2024)Cohere
1,239
1,227–1,250
$0.15 / $0.60
314 Granite 4.2 8BIBM
1,237
1,215–1,260
$0.06 / $0.15
315 OLMo 2 32B Instruct (0325)Ai2
1,234
1,214–1,253
–
316 Phi-4Microsoft
1,229
1,222–1,237
$0.07 / $0.14
317 Qwen2.5-Coder-32B-InstructAlibaba
1,228
1,213–1,243
$0.66 / $1
318 Ministral 8B (2410)Mistral AI
1,226
1,211–1,242
–
318 GPT-3.5 Turbo (0125)OpenAI
1,226
1,219–1,233
–
318 Mistral Large (2402)Mistral AI
1,226
1,218–1,234
–
321 Granite 4.2 3BIBM
1,223
1,199–1,248
–
321 Nova Micro 1.0Amazon
1,223
1,214–1,231
$0.035 / $0.14
323 Qwen1.5-110B-ChatAlibaba
1,219
1,209–1,228
–
324 Gemini Pro (Dev API)Google
1,218
1,206–1,229
–
325 Qwen1.5-72B-ChatAlibaba
1,214
1,206–1,223
–
325 Jamba 1.5 MiniAI21 Labs
1,214
1,202–1,227
–
327 Aya Expanse 8BCohere
1,213
1,202–1,224
–
328 Mixtral 8x22B InstructMistral AI
1,210
1,202–1,217
$2 / $6
329 Command RCohere
1,209
1,201–1,217
–
330 Hunyuan Standard 256KTencent
1,208
1,188–1,229
–
330 Gemini ProGoogle
1,208
1,189–1,227
–
330 Mistral MediumMistral AI
1,208
1,198–1,217
–
333 Reka Flash 21B (2024-02-26, online)Reka AI
1,204
1,192–1,216
–
334 Llama 3.1 Tülu 3 8BAi2
1,200
1,181–1,220
–
335 Reka Flash 21B (2024-02-26)Reka AI
1,198
1,188–1,208
–
336 Zephyr ORPO 141B A35B v0.1Hugging Face
1,196
1,178–1,214
–
337 WizardLM 70BMicrosoft
1,195
1,180–1,210
–
338 Gemma 2 2BGoogle
1,193
1,187–1,200
–
338 GPT-3.5 Turbo (1106)OpenAI
1,193
1,179–1,206
–
340 Llama 3 8B InstructMeta
1,192
1,185–1,199
–
341 Llama 3.1 8B InstructMeta
1,182
1,176–1,189
$0.05 / $0.08
342 Granite 3.1 8B InstructIBM
1,180
1,159–1,200
–
343 OpenChat 3.5OpenChat
1,177
1,161–1,192
–
344 DBRX Instruct PreviewDatabricks
1,176
1,167–1,186
–
345 Yi-1.5-34B-Chat01.AI
1,175
1,166–1,184
–
346 Phi-3-medium-4k-instructMicrosoft
1,173
1,164–1,182
–
347 Qwen1.5-32B-ChatAlibaba
1,172
1,162–1,183
–
348 Vicuna 33BLMSYS
1,171
1,161–1,182
–
348 Falcon 180B ChatTII
1,171
1,139–1,203
–
348 Mixtral 8x7B Instruct v0.1Mistral AI
1,171
1,163–1,178
–
348 DeepSeek LLM 67B ChatDeepSeek
1,171
1,152–1,189
–
352 Tülu 2 DPO 70BAi2
1,166
1,149–1,182
–
353 Qwen1.5-14B-ChatAlibaba
1,165
1,153–1,176
–
354 Yi-34B-Chat01.AI
1,164
1,153–1,176
–
354 OpenChat 3.5 (0106)OpenChat
1,164
1,151–1,176
–
356 Snowflake Arctic InstructSnowflake
1,160
1,150–1,170
–
357 OpenHermes 2.5 Mistral 7BTeknium
1,159
1,141–1,177
–
358 Gemma 1.1 7BGoogle
1,154
1,145–1,163
–
359 WizardLM 13BMicrosoft
1,153
1,138–1,168
–
360 InternLM2.5 20B ChatShanghai AI Lab
1,152
1,139–1,164
–
361 Vicuna 13BLMSYS
1,146
1,135–1,156
–
362 Granite 3.0 8B InstructIBM
1,144
1,129–1,160
–
363 Llama 3.2 3B InstructMeta
1,143
1,128–1,157
$0.05 / $0.33
364 Phi-3-small-8k-instructMicrosoft
1,141
1,131–1,151
–
364 Nous Hermes 2 Mixtral 8x7B DPONous Research
1,141
1,121–1,160
–
366 Starling-LM-7B-alphaUC Berkeley
1,140
1,127–1,153
–
366 MPT-30B-ChatMosaicML
1,140
1,117–1,162
–
366 Solar 10.7B Instruct v1.0Upstage
1,140
1,119–1,160
–
369 Dolphin 2.2.1 Mistral 7BCognitive Computations
1,132
1,103–1,160
–
370 Llama 2 70B ChatMeta
1,131
1,122–1,139
–
371 Llama 2 70B SteerLM ChatNVIDIA
1,128
1,106–1,150
–
371 Granite 3.1 2B InstructIBM
1,128
1,107–1,149
–
373 Starling-LM-7B-betaNexusflow
1,127
1,115–1,139
–
374 Guanaco 33BTim Dettmers
1,123
1,100–1,146
–
374 Zephyr 7B AlphaHugging Face
1,123
1,098–1,148
–
376 Qwen-14B-ChatAlibaba
1,121
1,103–1,140
–
376 Zephyr 7B BetaHugging Face
1,121
1,107–1,135
–
378 Vicuna 7BLMSYS
1,116
1,101–1,132
–
379 Mistral 7B Instruct v0.2Mistral AI
1,113
1,103–1,124
–
379 QwQ-32B-PreviewAlibaba
1,113
1,092–1,134
–
379 Qwen1.5-7B-ChatAlibaba
1,113
1,095–1,131
–
382 Granite 3.0 2B InstructIBM
1,106
1,091–1,122
–
383 Gemma 7BGoogle
1,105
1,091–1,120
–
383 Code Llama 34B InstructMeta
1,105
1,090–1,119
–
385 Llama 2 13B ChatMeta
1,104
1,094–1,115
–
385 Phi-3-mini-4k-instruct (June 2024)Microsoft
1,104
1,092–1,115
–
387 Phi-3-mini-128k-instructMicrosoft
1,099
1,087–1,111
–
388 PaLM 2Google
1,097
1,082–1,112
–
389 Mistral 7B InstructMistral AI
1,095
1,080–1,111
–
390 Phi-3-mini-4k-instructMicrosoft
1,093
1,082–1,103
–
390 StripedHyena Nous 7BTogether AI
1,093
1,075–1,111
–
392 Gemma 1.1 2BGoogle
1,092
1,079–1,105
–
393 Alpaca 13BStanford
1,081
1,062–1,100
–
394 SmolLM2 1.7B InstructHugging Face
1,079
1,053–1,106
–
395 Qwen1.5-4B-ChatAlibaba
1,077
1,062–1,092
–
396 Llama 3.2 1B InstructMeta
1,076
1,061–1,091
$0.027 / $0.20
396 Llama 2 7B ChatMeta
1,076
1,064–1,087
–
398 GPT4All 13B SnoozyNomic AI
1,062
1,031–1,093
–
399 Gemma 2BGoogle
1,058
1,039–1,077
–
400 MPT-7B-ChatMosaicML
1,057
1,037–1,078
–
401 Code Llama 70B InstructMeta
1,055
1,018–1,092
–
402 Koala 13BUC Berkeley
1,049
1,031–1,067
–
403 ChatGLM3-6BZ.ai
1,041
1,021–1,061
–
404 OLMo 7B InstructAi2
1,023
1,004–1,041
–
405 ChatGLM2-6BZ.ai
1,016
994–1,039
–
406 OpenAssistant Pythia 12BOpenAssistant
1,013
995–1,031
–
407 RWKV-4 Raven 14BRWKV
1,003
984–1,023
–
408 FastChat-T5 3BLMSYS
998
977–1,019
–
409 ChatGLM-6BZ.ai
982
961–1,002
–
410 Dolly v2 12BDatabricks
975
952–998
–
411 StableLM Tuned Alpha 7BStability AI
931
907–955
–
412 LLaMA 13BMeta
929
900–957
–

Swipe the table sideways for more columns.

Source: Arena (formerly LMArena), via Spring Prompt, springprompt.com/benchmarks/lmarena. Test your own prompt on these models →

Ranks follow the score as shown, so equal numbers share a rank. ≈ marks results whose 95% range overlaps the leader's: they cannot be told apart from it. Results as published by Arena (formerly LMArena); we do not re-run them.

Progress over time

Writing, literature and language, plotted at each model's release date, higher is better. Each model counts once, at its best setting. Scores are today's, not launch day's; 177 models without a known release date are left off. Hover or tap a mark for its model.

Best score so far

Every model at its release date. The line moves each time a new release beats every earlier one, and each new best is named where there is room.

1,0001,2001,4001,600 2024Jul2025Jul2026Jul GLM-4 (0520)Gemini 1.5 Pro (001)GPT-4o (2024-05-13)o1-previewo1GPT-4.5 PreviewClaude Sonnet 4.5Gemini 3 Pro PreviewClaude Opus 4.6Claude Opus 5.5Gemini 4 Argon

Best score by lab

Each lab's best so far, stepping whenever one of its releases beats its own earlier models. The 6 labs with the best scores today; 17 others not drawn.

GoogleAnthropicOpenAIMoonshot AIAlibabaMeta

1,2001,3001,4001,5001,600 2024Jul2025Jul2026Jul GoogleAnthropicOpenAIMoonshot AIAlibabaMeta

What it measures

Human preference on writing and language prompts.

What it does not measure

Not accuracy or correctness: it ranks which answer voters preferred, with answer length and formatting controlled for.

Contains data from the Arena Leaderboard Dataset by Arena, licensed under CC BY 4.0. Licence: Creative Commons Attribution 4.0 International.