Benchmarks / Arena (formerly LMArena)

Reported by Arena (formerly LMArena)

Arena (formerly LMArena)

Human preference on prompts Arena classes as expert-level.

Last updated 2 Oct 2026

Results dated
2 Oct 2026
Results
363 configurations of 338 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 prompts Arena classes as expert-level.

Latest results. , 338 models ranked.

Top 3 on Expert prompts (Arena rating, higher is better, each model at its best setting):

  1. Claude Sonnet 5.5 (extra-high reasoning) (Anthropic): 1,556
  2. Claude Fable 5 (high reasoning) (Anthropic): 1,549
  3. Claude Opus 4.6 (high reasoning) (Anthropic): 1,547

The first 22 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.

Expert prompts: GPT-4 Turbo

Top 15 of 338 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≈ Claude Sonnet 5.5 (extra-high reasoning)Anthropic 1,556
  2. 2≈ Claude Fable 5 (high reasoning)Anthropic 1,549
  3. 3≈ Claude Opus 4.6 (high reasoning)Anthropic 1,547
  4. 4≈ GPT-6.1 Sol (max reasoning)OpenAI 1,543
  5. 4≈ Claude Opus 5 (high reasoning)Anthropic 1,543
  6. 6≈ Claude Opus 5.5 (high reasoning)Anthropic 1,541
  7. 7≈ Gemini 4 Argon (high reasoning)Google 1,539
  8. 8≈ GPT-5.6 Sol (extra-high reasoning)OpenAI 1,535
  9. 9≈ Claude Opus 4.7 (high reasoning)Anthropic 1,534
  10. 10≈ Kimi K3 (max reasoning)Moonshot AI 1,533
  11. 11≈ MiMo-V2.6-ProXiaomi 1,532
  12. 12≈ Muse Spark 1.3 (max reasoning)Meta 1,529
  13. 13≈ Gemini 3.8 Flash (high reasoning)Google 1,526
  14. 14≈ Claude Opus 4.8 (high reasoning)Anthropic 1,525
  15. 15≈ Claude Fable 5.1 (max reasoning)Anthropic 1,521
  16. 238 GPT-4 TurboOpenAI 1,297

Full results

Arena Text: expert prompts, Arena rating, higher is better
#ModelExpert prompts · 95% range
Arena rating, higher is better
Price
$ per million tokens, in / out
1≈ Claude Sonnet 5.5 (extra-high reasoning)Anthropic
1,556
1,525–1,587
$2 / $10
2≈ Claude Fable 5 (high reasoning)Anthropic
1,549
1,540–1,559
$10 / $50
3≈ Claude Opus 4.6 (high reasoning)Anthropic · best of 2 settings
1,547
1,539–1,555
$5 / $25
4≈ GPT-6.1 Sol (max reasoning)OpenAI
1,543
1,511–1,576
$2 / $10
4≈ Claude Opus 5 (high reasoning)Anthropic · best of 2 settings
1,543
1,534–1,551
$5 / $25
6≈ Claude Opus 5.5 (high reasoning)Anthropic
1,541
1,514–1,567
$4 / $20
7≈ Gemini 4 Argon (high reasoning)Google
1,539
1,513–1,565
$2 / $10
8≈ GPT-5.6 Sol (extra-high reasoning)OpenAI
1,535
1,525–1,544
$4 / $20
9≈ Claude Opus 4.7 (high reasoning)Anthropic · best of 2 settings
1,534
1,526–1,542
$5 / $25
10≈ Kimi K3 (max reasoning)Moonshot AI
1,533
1,521–1,545
$3 / $15
11≈ MiMo-V2.6-ProXiaomi
1,532
1,504–1,559
$0.43 / $0.87
12≈ Muse Spark 1.3 (max reasoning)Meta
1,529
1,514–1,545
$1.25 / $4.25
13≈ Gemini 3.8 Flash (high reasoning)Google
1,526
1,515–1,537
$1.50 / $7.50
14≈ Claude Opus 4.8 (high reasoning)Anthropic · best of 2 settings
1,525
1,517–1,533
$5 / $25
15≈ Claude Fable 5.1 (max reasoning)Anthropic
1,521
1,502–1,540
$10 / $50
16≈ GLM-5.3 (max reasoning)Z.ai
1,518
1,504–1,532
$1.40 / $4.40
17≈ GPT-6 Astra (max reasoning)OpenAI
1,517
1,497–1,537
$10 / $50
17≈ DeepSeek-V4.1-Flash (max reasoning)DeepSeek
1,517
1,498–1,535
$0.30 / $1.20
19 GPT-5.4 (high reasoning)OpenAI · best of 2 settings
1,515
1,507–1,524
$2.50 / $15
20 Qwen3.8-Max (0902)Alibaba
1,513
1,501–1,525
$2 / $6
21 GPT-5.5 (high reasoning)OpenAI · best of 2 settings
1,512
1,505–1,520
$5 / $30
21 Claude Sonnet 5 (high reasoning)Anthropic
1,512
1,503–1,521
$2 / $10
23 Gemini 3.7 Flash (high reasoning)Google
1,511
1,499–1,523
$1.50 / $7.50
23 GLM-5.3-FlashZ.ai
1,511
1,498–1,523
$0.15 / $0.50
25≈ MiMo-V2.6-FlashXiaomi
1,509
1,487–1,531
$0.14 / $0.28
25 Gemini 3.1 Pro PreviewGoogle
1,509
1,502–1,515
$2 / $12
27 MiMo-V2.5-ProXiaomi
1,507
1,499–1,515
$0.43 / $0.87
28 Claude Sonnet 4.6Anthropic
1,506
1,498–1,514
$3 / $15
28≈ Qwen3.7-Max-PreviewAlibaba
1,506
1,475–1,537
–
30 GPT-5.6 Terra (extra-high reasoning)OpenAI
1,504
1,495–1,514
$2 / $12
30 Muse Spark 1.1Meta
1,504
1,494–1,513
$1.25 / $4.25
32 Kimi K2.6Moonshot AI
1,503
1,493–1,513
$0.95 / $4
32 Claude Opus 4.5 (high reasoning, 32k budget)Anthropic · best of 2 settings
1,503
1,491–1,515
$5 / $25
34≈ Muse Spark 1.2 (extra-high reasoning)Meta
1,502
1,476–1,528
$1.25 / $4.25
34 GLM-5.2 (max reasoning)Z.ai
1,502
1,493–1,511
$1.40 / $4.40
36 Qwen3.6-Max-PreviewAlibaba
1,500
1,475–1,525
–
36 Gemini 3.6 Flash (high reasoning)Google
1,500
1,490–1,509
$1.50 / $7.50
36 Qwen3.5-Max-PreviewAlibaba
1,500
1,487–1,512
–
39 Gemini 3 Pro PreviewGoogle
1,498
1,487–1,509
–
39 GPT-6 Luna (max reasoning)OpenAI
1,498
1,480–1,516
$0.10 / $0.50
39 GPT-6 Sol (max reasoning)OpenAI
1,498
1,478–1,517
$2 / $10
39 Gemma 4 31BGoogle
1,498
1,473–1,523
$0.15 / $0.40
39 Claude Sonnet 4.5 (high reasoning, 32k budget)Anthropic · best of 2 settings
1,498
1,490–1,506
$3 / $15
44 Gemini 3 Flash PreviewGoogle · best of 2 settings
1,496
1,484–1,509
$0.50 / $3
45 ERNIE 5.1Baidu
1,495
1,484–1,505
–
46 GLM-5.1Z.ai
1,494
1,486–1,503
$1.33 / $4.18
46≈ Nova Experimental Chat (2026-02-10)Amazon
1,494
1,462–1,525
–
48 MiMo-V2-ProXiaomi
1,493
1,480–1,506
–
49 Gemini 3.5 Flash (high reasoning)Google · best of 2 settings
1,492
1,484–1,501
$1.50 / $9
49 Muse SparkMeta
1,492
1,476–1,508
–
51 GPT-5.6 Luna (extra-high reasoning)OpenAI
1,491
1,482–1,501
$0.20 / $1.20
51 Grok 4.6 (high reasoning)xAI
1,491
1,479–1,503
$2 / $6
51 GPT-5.2 Chat (2026-02-10)OpenAI
1,491
1,480–1,502
–
54 Grok 4.5xAI
1,489
1,479–1,498
$2 / $6
55 Hy3Tencent
1,488
1,471–1,506
$0.14 / $0.58
56 Qwen3.7-PlusAlibaba
1,485
1,476–1,494
$0.32 / $1.28
57 DeepSeek-V4-Pro (0813, high reasoning)DeepSeek
1,484
1,466–1,503
$1.32 / $3.96
58 Kimi K2.5 (reasoning on)Moonshot AI · best of 2 settings
1,483
1,475–1,490
$0.57 / $2.85
59 Claude Opus 4.1 (16k reasoning budget)Anthropic · best of 2 settings
1,482
1,470–1,494
$15 / $75
59 GPT-5.1 (high reasoning)OpenAI · best of 2 settings
1,482
1,470–1,494
$1.25 / $10
59 GLM-5Z.ai
1,482
1,470–1,493
$1 / $3.20
62 Grok 4.20 Multi-Agent Beta (0309)xAI
1,481
1,473–1,489
–
63 GPT-5.4 mini (high reasoning)OpenAI
1,480
1,472–1,489
$0.75 / $4.50
63 DeepSeek-V4-Pro (0423)DeepSeek · best of 2 settings
1,480
1,471–1,488
$1.42 / $2.83
65 Grok 4.20 Beta (0309, reasoning)xAI
1,478
1,469–1,486
–
66 Dola-Seed-2.0-ProByteDance
1,477
1,470–1,485
–
66 Qwen3.5-397B-A17BAlibaba
1,477
1,470–1,484
$0.55 / $3.50
68 Qwen3.6-PlusAlibaba
1,476
1,467–1,485
$0.33 / $1.95
68 Grok 4.20 Beta 1xAI
1,476
1,463–1,488
–
68 Qwen3.8-27BAlibaba
1,476
1,462–1,489
$0.50 / $3
71 Gemma 4 26B A4BGoogle
1,475
1,450–1,501
$0.09 / $0.30
71 GPT-5.2 (high reasoning)OpenAI · best of 2 settings
1,475
1,465–1,485
$1.75 / $14
73 GPT-5.5 InstantOpenAI
1,473
1,461–1,485
–
73 InklingThinking Machines
1,473
1,463–1,483
$1 / $4.05
75 Nova Experimental Chat (2026-01-10)Amazon
1,472
1,438–1,505
–
75 Step 5 PreviewStepFun
1,472
1,438–1,505
$1 / $2.70
77 GPT-5.3 ChatOpenAI
1,471
1,459–1,482
–
78 Gemini 3.5 Flash-LiteGoogle
1,470
1,460–1,480
$0.30 / $2.50
78 MiMo-V2.5Xiaomi
1,470
1,461–1,479
$0.40 / $2
78 MiniMax-M3MiniMax
1,470
1,461–1,478
$0.30 / $1.20
81 LongCat-Flash-Chat (2602, experimental)Meituan
1,468
1,456–1,480
–
82 Nemotron 3 Ultra 550B A55B (NVFP4)NVIDIA
1,466
1,449–1,483
–
82 DeepSeek-V4-Flash (0423, high reasoning)DeepSeek · best of 2 settings
1,466
1,457–1,475
$0.05 / $0.45
82 Grok 4.1 ThinkingxAI
1,466
1,457–1,475
–
82 Qwen3-235B-A22B-Thinking-2507Alibaba
1,466
1,437–1,494
$0.45 / $3.50
86 Qwen3-Max-PreviewAlibaba
1,465
1,448–1,481
–
87 GLM-5V-TurboZ.ai
1,463
1,446–1,480
$1.20 / $4
88 MiMo-V2-OmniXiaomi
1,462
1,448–1,475
–
89 Kimi K2 Thinking TurboMoonshot AI
1,460
1,451–1,469
–
90 GPT-5 (high reasoning)OpenAI
1,459
1,444–1,475
$1.25 / $10
91 Gemini 2.5 ProGoogle
1,458
1,452–1,465
$1.25 / $10
92 Grok 4.7 (extra-high reasoning)xAI
1,457
1,434–1,480
$2 / $6
93 MiniMax-M2.7MiniMax
1,453
1,445–1,460
$0.30 / $1.20
94 DeepSeek-V3.2 (reasoning on)DeepSeek · best of 2 settings
1,452
1,441–1,464
$0.28 / $0.42
94 Grok 4.1xAI
1,452
1,444–1,461
–
94 Grok 4.3xAI
1,452
1,444–1,460
$1.25 / $2.50
97 Claude Haiku 4.5Anthropic
1,451
1,445–1,457
$1 / $5
97 ERNIE 5.0 Preview (1203)Baidu
1,451
1,429–1,472
–
99 DeepSeek-V3.2-Exp (reasoning on)DeepSeek · best of 2 settings
1,449
1,420–1,477
$0.27 / $0.41
99 GLM-4.7Z.ai
1,449
1,428–1,469
$0.60 / $2.20
99 o3OpenAI
1,449
1,437–1,460
$2 / $8
102 Claude Opus 4 (16k reasoning budget)Anthropic · best of 2 settings
1,448
1,434–1,462
–
103 Qwen3-235B-A22B-Instruct-2507Alibaba
1,447
1,440–1,455
$0.15 / $0.75
103 ERNIE 5.0 (0110)Baidu
1,447
1,436–1,458
–
105 Hy3 PreviewTencent
1,446
1,424–1,468
$0.18 / $0.60
106 Inkling SmallThinking Machines
1,445
1,434–1,456
$0.45 / $1.20
107 GLM-4.5Z.ai
1,444
1,426–1,461
$0.60 / $2.20
108 Gemini 3.1 Flash-Lite PreviewGoogle
1,443
1,435–1,452
$0.25 / $1.50
108 GPT-5 ChatOpenAI
1,443
1,428–1,458
–
110 Muse GlimmerMeta
1,442
1,414–1,470
–
110 Qwen3.5-122B-A10BAlibaba
1,442
1,430–1,453
$0.26 / $2.08
112 Qwen3.5-27BAlibaba
1,441
1,429–1,453
$0.27 / $2.16
113 Qwen3-VL-235B-A22B-InstructAlibaba
1,440
1,415–1,465
$0.21 / $1.90
114 Qwen3-VL-235B-A22B-ThinkingAlibaba
1,439
1,409–1,468
–
114 GLM-4.6Z.ai
1,439
1,426–1,451
$0.50 / $2
116 Grok 4.1 Fast (reasoning)xAI
1,438
1,429–1,447
–
117 GPT-5.4 nano (high reasoning)OpenAI
1,437
1,429–1,446
$0.20 / $1.25
118 Solar Pro 4Upstage
1,436
1,422–1,451
$0.15 / $0.60
119 Mistral Medium 3.5Mistral AI
1,435
1,417–1,452
$1.50 / $7.50
120 Claude Sonnet 4 (32k reasoning budget)Anthropic · best of 2 settings
1,433
1,418–1,447
–
121 ERNIE 5.0 Preview (1022)Baidu
1,432
1,399–1,466
–
121 DeepSeek-V3.1 (reasoning on)DeepSeek · best of 2 settings
1,432
1,407–1,457
$0.27 / $1
121 Gemini 2.5 Flash Preview (09-2025)Google
1,432
1,418–1,446
–
124 Grok 4 (0709)xAI
1,431
1,418–1,444
–
125 Step 3.5 FlashStepFun
1,429
1,420–1,437
$0.10 / $0.30
125 ChatGPT-4o (2025-03-26)OpenAI
1,429
1,420–1,438
–
125 GPT-4.5 PreviewOpenAI
1,429
1,405–1,452
–
128 LongCat-Flash-ChatMeituan
1,428
1,402–1,454
–
129 Mistral Large 3Mistral AI
1,427
1,420–1,434
$0.50 / $1.50
130 MiniMax-M2.5MiniMax
1,425
1,415–1,435
$0.29 / $1.20
131 Gemini 2.5 FlashGoogle
1,424
1,417–1,430
$0.30 / $2.50
132 Grok 4 Fast (reasoning)xAI
1,421
1,401–1,440
–
133 MiniMax-M2.1 PreviewMiniMax
1,420
1,403–1,437
–
134 Qwen3.5-35B-A3BAlibaba
1,419
1,408–1,431
$0.15 / $1
134 Qwen3.5-FlashAlibaba
1,419
1,411–1,428
$0.065 / $0.26
136 DeepSeek-R1-0528DeepSeek
1,418
1,399–1,438
$0.50 / $2.18
137 Qwen3-MaxAlibaba
1,417
1,391–1,444
–
137 Kimi K2 (0905)Moonshot AI
1,417
1,393–1,441
$0.60 / $2.50
137 MiMo-V2-Flash (no reasoning)Xiaomi · best of 2 settings
1,417
1,407–1,427
–
140 Grok 4 Fast ChatxAI
1,416
1,382–1,451
–
140 Kimi K2 (0711)Moonshot AI
1,416
1,400–1,432
$0.57 / $2.30
142 Claude 3.7 Sonnet (32k reasoning budget)Anthropic · best of 2 settings
1,412
1,398–1,426
–
143 Mistral Medium 3.1Mistral AI
1,410
1,402–1,418
$0.40 / $2
144 o4-miniOpenAI
1,408
1,395–1,420
–
145 Hunyuan T1 (2025-07-11)Tencent
1,407
1,368–1,445
–
145 Grok 3 Mini (high reasoning)xAI
1,407
1,388–1,426
–
147 GPT-4.1OpenAI
1,405
1,393–1,417
$2 / $8
147 Nova Experimental Chat (11-10)Amazon
1,405
1,390–1,419
–
149 Trinity Large PreviewArcee AI
1,404
1,392–1,415
–
150 Grok 3 Preview (02-24)xAI
1,403
1,388–1,418
–
151 DeepSeek-R1DeepSeek
1,402
1,382–1,422
–
151 o1OpenAI
1,402
1,385–1,419
–
153 GPT-5 mini (high reasoning)OpenAI
1,401
1,383–1,418
$0.25 / $2
154 Trinity Large ThinkingArcee AI
1,400
1,388–1,412
–
154 Qwen3-Next-80B-A3B-InstructAlibaba
1,400
1,382–1,417
$0.09 / $1.10
156 GLM-4.5VZ.ai
1,399
1,357–1,441
$0.60 / $1.80
157 o3-mini (high reasoning)OpenAI · best of 2 settings
1,398
1,378–1,419
–
157 DeepSeek-V3-0324DeepSeek
1,398
1,386–1,411
$0.25 / $1
157 Qwen3-32BAlibaba
1,398
1,360–1,435
$0.14 / $0.40
160 Nova Experimental Chat (12-10)Amazon
1,396
1,362–1,431
–
161 Qwen3-235B-A22B (no reasoning)Alibaba · best of 2 settings
1,395
1,382–1,409
–
161 Granite 4.2 30BIBM
1,395
1,364–1,426
–
161 Qwen3-30B-A3B-Instruct-2507Alibaba
1,395
1,378–1,412
$0.09 / $0.30
164 GLM-4.6VZ.ai
1,394
1,351–1,437
$0.30 / $0.90
165 Qwen3-Next-80B-A3B-ThinkingAlibaba
1,393
1,370–1,416
–
165 Nemotron 3 Super 120B A12BNVIDIA
1,393
1,370–1,415
$0.08 / $0.45
167 GLM-4.5-AirZ.ai
1,389
1,374–1,405
$0.14 / $0.86
168 Nemotron 3.5 Lightning 30B A3B (NVFP4)NVIDIA
1,388
1,373–1,404
–
169 GLM-4.7-FlashZ.ai
1,386
1,367–1,406
$0.06 / $0.40
169 Gemini 2.5 Flash-Lite Preview (09-2025, no reasoning)Google
1,386
1,375–1,397
–
171 GPT-4.1 miniOpenAI
1,383
1,370–1,396
$0.40 / $1.60
171 o1-previewOpenAI
1,383
1,368–1,397
–
173 Grok 3 Mini BetaxAI
1,380
1,362–1,397
–
174 Qwen3-Coder-480B-A35BAlibaba
1,379
1,362–1,395
$0.30 / $1
174 Mistral Medium 3Mistral AI
1,379
1,365–1,392
$0.40 / $2
176 Claude 3.5 Sonnet (2024-10-22)Anthropic
1,372
1,362–1,381
–
177 Llama 3.3 Nemotron Super 49B v1.5NVIDIA
1,371
1,330–1,413
–
177 Gemini 2.5 Flash-Lite Preview (06-17, reasoning on)Google
1,371
1,356–1,386
–
177 Granite 4.2 8BIBM
1,371
1,337–1,405
$0.06 / $0.15
180 Qwen-PlusAlibaba
1,370
1,340–1,400
$0.26 / $0.78
181 Qwen2.5-MaxAlibaba
1,368
1,354–1,382
–
182 Ring-flash-2.0Ant Group
1,367
1,335–1,399
–
183 MiniMax-M1MiniMax
1,365
1,351–1,380
–
183 Nova Experimental Chat (10-20)Amazon
1,365
1,343–1,386
–
185 Step 3StepFun
1,363
1,328–1,399
–
186 Ling-flash-2.0Ant Group
1,362
1,332–1,392
–
187 QwQ-32BAlibaba
1,361
1,344–1,377
–
188 gpt-oss-120bOpenAI
1,357
1,341–1,373
$0.10 / $0.50
189 Gemini 2.0 Flash (001)Google
1,356
1,343–1,369
–
190 INTELLECT-3Prime Intellect
1,354
1,323–1,386
–
190 GPT-5 nano (high reasoning)OpenAI
1,354
1,321–1,386
$0.05 / $0.40
192 Nova 2 LiteAmazon
1,352
1,332–1,373
$0.30 / $2.50
193 o1-miniOpenAI
1,351
1,340–1,363
–
194 MiniMax-M2MiniMax
1,350
1,317–1,382
–
195 Hunyuan TurboS (2025-04-16)Tencent
1,349
1,325–1,373
–
196 Mercury 2Inception
1,348
1,312–1,384
–
196 DeepSeek-V3DeepSeek
1,348
1,331–1,364
$0.26 / $1.03
198 Claude 3.5 Sonnet (2024-06-20)Anthropic
1,343
1,332–1,355
–
198 Command ACohere
1,343
1,332–1,355
$2.50 / $10
200 Qwen3-30B-A3BAlibaba
1,342
1,326–1,359
$0.12 / $0.50
201 Gemini 1.5 Pro (002)Google
1,339
1,328–1,350
–
201 Granite 4.1 8BIBM
1,339
1,307–1,370
–
203 Gemma 3 27BGoogle
1,337
1,324–1,350
$0.08 / $0.35
204 Olmo 3.1 32B ThinkAi2
1,335
1,310–1,361
–
205 Mistral Small 3.2 24BMistral AI
1,334
1,315–1,354
$0.094 / $0.25
206 Gemini 2.0 Flash-Lite Preview (02-05)Google
1,332
1,315–1,348
–
207 Yi-Lightning01.AI
1,331
1,316–1,345
–
207 Nemotron 3 Nano 30B A3BNVIDIA
1,331
1,312–1,350
$0.06 / $0.24
209 Granite 4.2 3BIBM
1,330
1,295–1,366
–
210 Olmo 3.1 32B InstructAi2
1,326
1,304–1,348
–
210 Llama 4 MaverickMeta
1,326
1,312–1,339
$0.35 / $1
210 Step-2 16K Exp (2024-12)StepFun
1,326
1,294–1,357
–
213 Llama 3.1 405B Instruct (FP8)Meta
1,325
1,313–1,336
–
214 Hunyuan Large (2025-02-10)Tencent
1,323
1,286–1,359
–
215 Qwen2.5-Plus (1127)Alibaba
1,321
1,300–1,343
–
216 Gemini 1.5 Pro (001)Google
1,320
1,308–1,332
–
217 Claude 3 OpusAnthropic
1,318
1,309–1,328
–
218 Claude 3.5 HaikuAnthropic
1,316
1,305–1,326
–
218 Grok 2 (2024-08-13)xAI
1,316
1,305–1,327
–
220 gpt-oss-20bOpenAI
1,314
1,287–1,342
$0.03 / $0.14
221 GPT-4.1 nanoOpenAI
1,313
1,283–1,344
$0.10 / $0.40
222 Granite H SmallIBM
1,312
1,276–1,348
–
222 Llama 3.1 405B Instruct (BF16)Meta
1,312
1,299–1,325
–
222 GLM-4-Plus (0111)Z.ai
1,312
1,282–1,341
–
225 GPT-4o (2024-08-06)OpenAI
1,311
1,299–1,324
$2.50 / $10
225 GPT-4o (2024-05-13)OpenAI
1,311
1,301–1,321
$5 / $15
225 Athene-V2-ChatNexusflow
1,311
1,296–1,326
–
228 Llama 4 ScoutMeta
1,308
1,293–1,324
$0.10 / $0.30
229 DeepSeek-V2.5-1210DeepSeek
1,306
1,280–1,332
–
230 GLM-4-PlusZ.ai
1,303
1,288–1,318
–
230 Olmo 3 32B ThinkAi2
1,303
1,267–1,339
–
232 Mistral Small 3.1 24BMistral AI
1,302
1,287–1,317
$0.35 / $0.56
232 Mistral Large 2 (2407)Mistral AI
1,302
1,289–1,314
$2 / $6
234 Athene 70B (0725)Nexusflow
1,301
1,282–1,320
–
235 Qwen-Max (0919)Alibaba
1,300
1,282–1,318
–
236 Gemini Advanced (0514)Google
1,298
1,284–1,313
–
236 Llama 3.3 70B InstructMeta
1,298
1,286–1,309
$0.71 / $0.71
238 Hunyuan Large VisionTencent
1,297
1,264–1,330
–
238 GPT-4 TurboOpenAI
1,297
1,286–1,308
–
238 Step-1o Turbo (2025-06)StepFun
1,297
1,270–1,324
–
241 Llama 3.1 Nemotron 70B InstructNVIDIA
1,296
1,269–1,323
–
241 DeepSeek-V2.5DeepSeek
1,296
1,281–1,311
–
243 Grok 2 Mini (2024-08-13)xAI
1,295
1,283–1,307
–
243 Qwen2.5-72B-InstructAlibaba
1,295
1,283–1,307
$0.36 / $0.40
245 Magistral Medium 1.0Mistral AI
1,293
1,266–1,320
–
246 Reka Core (2024-09-04)Reka AI
1,291
1,266–1,316
–
247 GPT-4 Turbo Preview (1106)OpenAI
1,290
1,278–1,302
–
247 Hunyuan Standard (2025-02-10)Tencent
1,290
1,251–1,328
–
249 Jamba 1.5 LargeAI21 Labs
1,285
1,256–1,314
–
249 GPT-4o mini (2024-07-18)OpenAI
1,285
1,274–1,296
$0.15 / $0.60
249 Gemini 1.5 Flash (002)Google
1,285
1,272–1,298
–
252 GPT-4 Turbo Preview (0125)OpenAI
1,280
1,268–1,292
–
253 Gemma 3n E4BGoogle
1,279
1,261–1,298
–
254 Reka Flash (2024-09-04)Reka AI
1,277
1,254–1,300
–
255 Qwen2.5-Coder-32B-InstructAlibaba
1,276
1,244–1,309
$0.66 / $1
256 Llama 3.1 70B InstructMeta
1,275
1,264–1,287
$0.72 / $0.72
257 Gemma 3 12BGoogle
1,274
1,232–1,315
$0.05 / $0.15
258 Nova Pro 1.0Amazon
1,273
1,257–1,289
$0.80 / $3.20
258 Claude 3 SonnetAnthropic
1,273
1,261–1,285
–
260 Mistral Large 2.1 (2411)Mistral AI
1,271
1,256–1,286
–
261 Phi-4Microsoft
1,268
1,251–1,285
$0.07 / $0.14
261 Gemini 1.5 Flash (001)Google
1,268
1,256–1,281
–
263 Aya Expanse 32BCohere
1,267
1,253–1,281
–
264 DeepSeek-Coder-V2DeepSeek
1,266
1,245–1,288
–
265 GPT-4 (0314)OpenAI
1,264
1,249–1,280
–
266 Nova Lite 1.0Amazon
1,262
1,245–1,279
$0.06 / $0.24
267 Qwen2-72B-InstructAlibaba
1,261
1,246–1,275
–
268 Nemotron-4 340B InstructNVIDIA
1,260
1,242–1,279
–
268 Gemma 2 27BGoogle
1,260
1,250–1,271
$0.65 / $0.65
268 Gemma 3 4BGoogle
1,260
1,219–1,301
$0.04 / $0.05
271 Mistral Small 3Mistral AI
1,259
1,238–1,279
$0.05 / $0.08
272 GPT-4 (0613)OpenAI
1,253
1,240–1,266
–
272 Claude 3 HaikuAnthropic
1,253
1,242–1,264
–
272 Llama 3.1 Nemotron 51B InstructNVIDIA
1,253
1,220–1,286
–
275 GLM-4 (0520)Z.ai
1,252
1,227–1,277
–
276 Command R+ (08-2024)Cohere
1,250
1,225–1,274
$2.50 / $10
277 Llama 3 70B InstructMeta
1,245
1,234–1,256
–
278 Nova Micro 1.0Amazon
1,244
1,227–1,261
$0.035 / $0.14
279 Gemini 1.5 Flash-8B (001)Google
1,243
1,230–1,256
–
280 Command R+Cohere
1,242
1,229–1,254
–
281 Gemma 2 9BGoogle
1,238
1,226–1,249
–
282 Ministral 8B (2410)Mistral AI
1,237
1,207–1,268
–
283 Qwen1.5-72B-ChatAlibaba
1,236
1,220–1,251
–
284 Granite 3.1 8B InstructIBM
1,235
1,200–1,271
–
284 Gemma 2 9B SimPOPrinceton NLP
1,235
1,206–1,265
–
286 Aya Expanse 8BCohere
1,232
1,208–1,256
–
287 Command R (08-2024)Cohere
1,231
1,209–1,253
$0.15 / $0.60
288 Mistral Large (2402)Mistral AI
1,229
1,215–1,243
–
288 Qwen1.5-110B-ChatAlibaba
1,229
1,212–1,245
–
290 Mistral MediumMistral AI
1,226
1,209–1,243
–
290 Reka Flash 21B (2024-02-26, online)Reka AI
1,226
1,205–1,247
–
292 Qwen1.5-32B-ChatAlibaba
1,223
1,205–1,242
–
293 Yi-1.5-34B-Chat01.AI
1,221
1,202–1,239
–
294 InternLM2.5 20B ChatShanghai AI Lab
1,220
1,198–1,242
–
295 Mixtral 8x22B InstructMistral AI
1,218
1,204–1,232
$2 / $6
295 Granite 3.1 2B InstructIBM
1,218
1,183–1,254
–
297 Reka Flash 21B (2024-02-26)Reka AI
1,216
1,199–1,233
–
298 GPT-3.5 Turbo (1106)OpenAI
1,215
1,187–1,243
–
298 Command RCohere
1,215
1,201–1,228
–
300 Jamba 1.5 MiniAI21 Labs
1,213
1,182–1,244
–
301 Llama 3 8B InstructMeta
1,211
1,199–1,223
–
302 Phi-3-medium-4k-instructMicrosoft
1,205
1,187–1,223
–
302 Granite 3.0 8B InstructIBM
1,205
1,174–1,236
–
304 Llama 3.1 8B InstructMeta
1,198
1,185–1,210
$0.05 / $0.08
305 GPT-3.5 Turbo (0125)OpenAI
1,197
1,185–1,210
–
305 Mixtral 8x7B Instruct v0.1Mistral AI
1,197
1,184–1,210
–
307 Qwen1.5-14B-ChatAlibaba
1,194
1,174–1,214
–
308 DBRX Instruct PreviewDatabricks
1,191
1,174–1,207
–
309 Gemini Pro (Dev API)Google
1,180
1,156–1,203
–
310 Llama 3.2 3B InstructMeta
1,179
1,154–1,203
$0.05 / $0.33
311 Granite 3.0 2B InstructIBM
1,177
1,149–1,206
–
312 Gemma 1.1 7BGoogle
1,175
1,157–1,192
–
313 Starling-LM-7B-betaNexusflow
1,174
1,154–1,194
–
314 Zephyr ORPO 141B A35B v0.1Hugging Face
1,172
1,133–1,212
–
314 Gemma 2 2BGoogle
1,172
1,159–1,185
–
316 Phi-3-small-8k-instructMicrosoft
1,167
1,147–1,186
–
317 Snowflake Arctic InstructSnowflake
1,165
1,148–1,182
–
318 OpenChat 3.5OpenChat
1,164
1,123–1,205
–
318 OpenChat 3.5 (0106)OpenChat
1,164
1,141–1,187
–
320 Yi-34B-Chat01.AI
1,156
1,131–1,181
–
321 QwQ-32B-PreviewAlibaba
1,153
1,118–1,189
–
322 Phi-3-mini-4k-instruct (June 2024)Microsoft
1,149
1,123–1,175
–
323 Qwen1.5-7B-ChatAlibaba
1,148
1,112–1,185
–
324 Vicuna 33BLMSYS
1,146
1,120–1,173
–
325 Phi-3-mini-4k-instructMicrosoft
1,140
1,121–1,160
–
325 Mistral 7B Instruct v0.2Mistral AI
1,140
1,120–1,161
–
327 Starling-LM-7B-alphaUC Berkeley
1,138
1,107–1,168
–
328 Llama 2 70B ChatMeta
1,137
1,120–1,155
–
329 Llama 2 7B ChatMeta
1,134
1,106–1,162
–
330 Llama 2 13B ChatMeta
1,130
1,104–1,155
–
331 Gemma 7BGoogle
1,122
1,090–1,153
–
332 Gemma 1.1 2BGoogle
1,117
1,092–1,143
–
332 Vicuna 13BLMSYS
1,117
1,086–1,148
–
334 Zephyr 7B BetaHugging Face
1,113
1,075–1,151
–
335 Qwen1.5-4B-ChatAlibaba
1,108
1,077–1,139
–
336 Phi-3-mini-128k-instructMicrosoft
1,098
1,078–1,119
–
337 Mistral 7B InstructMistral AI
1,085
1,046–1,124
–
338 Llama 3.2 1B InstructMeta
1,082
1,054–1,110
$0.027 / $0.20

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. Each model is shown at its best setting; show every setting. Results as published by Arena (formerly LMArena); we do not re-run them.

Progress over time

Expert prompts, 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; 130 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)Claude 3 OpusGemini 1.5 Pro (001)Claude 3.5 Sonnet (2024-06-20)o1-previewo1Claude 3.7 SonnetGPT-4.5 PreviewGemini 2.5 ProQwen3-235B-A22B-Thinking-2507Gemini 3 Pro PreviewClaude Opus 4.5Claude Opus 4.6Claude Fable 5Claude Sonnet 5.5

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.

AnthropicOpenAIGoogleMoonshot AIXiaomiMeta

1,2001,3001,4001,5001,600 2024Jul2025Jul2026Jul AnthropicOpenAIGoogleMoonshot AIXiaomiMeta

What it measures

Human preference on prompts Arena classes as expert-level.

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.