Gemini 4 Argon1526
Claude Opus 5.51516Claude Fable 5.11513Claude Opus 51508Claude Opus 4.61502▲4
Muse Wonder1501Gemini 3.8 Flash1498
MiMo-V2.6-Pro1491Muse Spark 1.31491Claude Fable 51491▼2Claude Opus 4.71490▲6Gemini 3.7 Flash1490
Qwen3.8-Max1484▲3Gemini 3.5 Flash1482▼1Muse Spark 1.21482▼8Gemini 3.1 Pro Preview1480▲0Gemini 3.6 Flash1480▲0Gemini 3 Pro1480Muse Spark 1.11478▼2Qwen3.7-Max-Preview1475▲1
Kimi K31475▲2Muse Spark1474▲1
GLM-5.31474Claude Sonnet 5.51473GLM-5.3-Flash1472▲1
GPT-5.51472▲6Qwen3.5-Max-Preview1470▼1GLM-5.21470▲6GPT-5.41469▲17
ERNIE 5.11468▼0Gemini 3 Flash1466▼1
DeepSeek-V4.1-Flash-202609101464Gemini 4 Argon1526Claude Opus 5.51516Claude Fable 5.11513Claude Opus 51508Claude Opus 4.61502▲4Muse Wonder1501Gemini 3.8 Flash1498MiMo-V2.6-Pro1491Muse Spark 1.31491Claude Fable 51491▼2Claude Opus 4.71490▲6Gemini 3.7 Flash1490Qwen3.8-Max1484▲3Gemini 3.5 Flash1482▼1Muse Spark 1.21482▼8Gemini 3.1 Pro Preview1480▲0Gemini 3.6 Flash1480▲0Gemini 3 Pro1480Muse Spark 1.11478▼2Qwen3.7-Max-Preview1475▲1Kimi K31475▲2Muse Spark1474▲1GLM-5.31474Claude Sonnet 5.51473GLM-5.3-Flash1472▲1GPT-5.51472▲6Qwen3.5-Max-Preview1470▼1GLM-5.21470▲6GPT-5.41469▲17ERNIE 5.11468▼0Gemini 3 Flash1466▼1DeepSeek-V4.1-Flash-202609101464MiMo-V2.5-Pro1464▼1GLM-5.11463▲0Claude Opus 4.81461▲8Claude Sonnet 4.61458▼1Gemini 2.5 Pro1458GPT-5.6 Sol1457Kimi K2.61455▲0Qwen3.7-Plus1455▲0MiMo-V2.6-Flash1452DeepSeek-V4-Pro1451▲1Claude Opus 4.51451▲0
Grok 4.20 Beta1451Grok 4.20 Multi-Agent Beta1451▲1
amazon-nova-experimental-chat1449
Dola Seed 2.0 Pro1448▲1GPT-5.6 Terra1448Grok 4.51448▼2Qwen3.6-Max-Preview1447▲1GLM-51446▼0Kimi K2.51445
NVIDIA-Nemotron-3-Ultra-550B-A55B1445ERNIE 5.01445▲0Grok 4.20 Beta 11445▲0GPT-6.1 Sol1444
Gemma 4 31B1443▲1GPT-5.11443▲20Claude Sonnet 51442GPT-6 Astra1442IInkling1442▲2
Step 5 Preview1442GLM-4.61440▼0Qwen3.8-27B1440▲1MiMo-V2.5-Pro1464▼1GLM-5.11463▲0Claude Opus 4.81461▲8Claude Sonnet 4.61458▼1Gemini 2.5 Pro1458GPT-5.6 Sol1457Kimi K2.61455▲0Qwen3.7-Plus1455▲0MiMo-V2.6-Flash1452DeepSeek-V4-Pro1451▲1Claude Opus 4.51451▲0Grok 4.20 Beta1451Grok 4.20 Multi-Agent Beta1451▲1amazon-nova-experimental-chat1449Dola Seed 2.0 Pro1448▲1GPT-5.6 Terra1448Grok 4.51448▼2Qwen3.6-Max-Preview1447▲1GLM-51446▼0Kimi K2.51445NVIDIA-Nemotron-3-Ultra-550B-A55B1445ERNIE 5.01445▲0Grok 4.20 Beta 11445▲0GPT-6.1 Sol1444Gemma 4 31B1443▲1GPT-5.11443▲20Claude Sonnet 51442GPT-6 Astra1442IInkling1442▲2Step 5 Preview1442GLM-4.61440▼0Qwen3.8-27B1440▲1Explorer
Every AI model, ranked by people’s votes and by benchmarks, with its price and context window. Gemini 4 Argon leads LMArena’s text leaderboard with a rating of 1526.
- Models
- 440
- Labs
- 42
- Leaderboards
- 37
- Arena votes
- 10.2M
- Updated
- Oct 24, 2026
All 440 models, by lab
Alibaba49
- Qwen3.8-Max
- Qwen3.7-Max-Preview
- Qwen3.5-Max-Preview
- Qwen3.7-Plus
- Qwen3.6-Max-Preview
- Qwen3.8-27B
- Qwen3-Max-Preview
- Qwen3.5-397B-A17B
- Qwen3.6-Plus
- Qwen3-VL-235B-A22B-Instruct
- Qwen3-235B-A22B-Instruct
- Qwen3.5-122B-A10B
- Qwen3-Next-80B-A3B-Instruct
- Qwen3-235B-A22B-Thinking
- Qwen3.5-27B
- Qwen3-VL-235B-A22B-Thinking
- Qwen3.5-Flash
- Qwen3.5-35B-A3B
- Qwen3-235B-A22B
- Qwen3-30B-A3B-Instruct
- Qwen3-Next-80B-A3B-Thinking
- Qwen2.5-Max
- Qwen3-Coder-480B-A35B-Instruct
- Qwen3-32B
- QwQ-32B
- Qwen-Plus
- Qwen3-30B-A3B
- Qwen2.5-Plus
- Qwen-Max
- Qwen2.5-72B-Instruct
- Qwen2.5-Coder-32B-Instruct
- Qwen2-72B-Instruct
- Qwen1.5-110B-Chat
- Qwen1.5-72B-Chat
- Qwen1.5-32B-Chat
- Qwen1.5-14B-Chat
- Qwen1.5-7B-Chat
- Qwen-14B-Chat
- Qwen1.5-4B-Chat
- Qwen 2.5 Instruct Turbo
- Qwen 3 VL Plus
- Qwen 3.5 Plus
- Qwen 3.6 27B
- Qwen-VL-Max
- Qwen2-VL-72B
- Qwen2-VL-7B-Instruct
- Qwen2.5-VL-32B-Instruct
- Qwen2.5-VL-72B-Instruct
- Qwen3.8-Flash-Next
OpenAI44
- GPT-5.5
- GPT-5.4
- GPT-5.6 Sol
- GPT-5.6 Terra
- GPT-6.1 Sol
- GPT-5.1
- GPT-6 Astra
- GPT-5.2 Chat
- GPT-5.6 Luna
- ChatGPT-4o Latest
- GPT-5.5 Instant
- GPT-4.5 Preview
- GPT-5.2
- GPT-5.4 mini
- o3
- GPT-5
- GPT-5 Chat
- GPT-6 Sol
- GPT-6 Luna
- GPT-5.3 Chat
- GPT-4.1
- GPT-5 mini
- GPT-5.4 nano
- o1
- gpt-oss-120b
- o4-mini
- GPT-4.1 mini
- o3-mini
- GPT-5 nano
- o1-mini
- GPT-4o
- gpt-oss-20b
- GPT-4o mini
- GPT-4.1 nano
- GPT-4 Turbo
- GPT-4
- GPT-3.5 Turbo
- GPT4All 13B Snoozy
- GPT 5 Codex
- GPT 5.1 Codex Max
- GPT-5.1-Codex
- GPT-5.1-Codex-Mini
- GPT-5.2-Codex
- GPT-5.3-Codex
Google37
- Gemini 4 Argon
- Gemini 3.8 Flash
- Gemini 3.7 Flash
- Gemini 3.5 Flash
- Gemini 3.1 Pro Preview
- Gemini 3.6 Flash
- Gemini 3 Pro
- Gemini 3 Flash
- Gemini 2.5 Pro
- Gemma 4 31B
- Gemini 3.5 Flash-Lite
- Gemma 4 26B A4B
- Gemini 2.5 Flash
- Gemini 3.1 Flash-Lite Preview
- Gemini 2.5 Flash-Lite Preview
- Gemma 3 27B IT
- Gemini 2.0 Flash
- Gemma 3 12B IT
- Gemini 2.0 Flash-Lite Preview
- Gemini 1.5 Pro
- Gemma 3n E4B IT
- Gemma 3 4B IT
- Gemini 1.5 Flash
- Gemini Advanced
- Gemma 2 27B IT
- Gemma 2 9B IT SimPO
- Gemini 1.5 Flash-8B
- Gemma 2 9B IT
- Gemma 2 2B IT
- Gemini Pro
- Gemma 1.1 7B IT
- Gemma 7B IT
- PaLM 2
- Gemma 1.1 2B IT
- Gemma 2B IT
- Gemini 2.5 Flash Preview
- Gemini 2.5 Flash Preview 4/17
Meta32
- Muse Wonder
- Muse Spark 1.3
- Muse Spark 1.2
- Muse Spark 1.1
- Muse Spark
- Muse Glimmer
- Llama 4 Maverick 17B 128E Instruct
- Llama 3.1 405B Instruct
- Llama 4 Scout 17B 16E Instruct
- Llama 3.3 70B Instruct
- Llama 3.1 70B Instruct
- Llama 3 70B Instruct
- Llama 3.1 8B Instruct
- Llama 3 8B Instruct
- Llama 2 70B Chat
- Llama 3.2 3B Instruct
- Llama 2 13B Chat
- Code Llama 34B Instruct
- Code Llama 70B Instruct
- Llama 3.2 1B Instruct
- Llama 2 7B Chat
- LLaMA 13B
- Llama 3.1 Instruct Turbo
- Llama 3.2 11B Vision Instruct
- Llama 3.2 90B Vision Instruct
- Llama 3.2 Vision
- Llama 3.3 Instruct Turbo
- Llama 3.3 Nemotron Super
- Llama 4 Scout
- Muse Spark 1.3 Max
- Muse Spark Open
- Muse Wonder Full
Mistral AI27
- Mistral Large 4
- Mistral Large 3
- Mistral Medium 3.1
- Mistral Medium 3.5
- Mistral Medium 3
- Mistral Small 3.2
- Mistral Small 3.1 24B Instruct
- Mistral Large 2
- Mistral Large
- Magistral Medium
- Mistral Small 24B Instruct
- Ministral 8B
- Mistral Medium
- Mixtral 8x22B Instruct v0.1
- Mixtral 8x7B Instruct v0.1
- OpenHermes 2.5 Mistral 7B
- Mistral 7B Instruct v0.2
- Dolphin 2.2.1 Mistral 7B
- Mistral 7B Instruct
- Devstral 2
- Devstral Medium
- Devstral Small 2
- Magistral Small 1.2
- Mistral Small
- Mistral Small 4
- Pixtral 12B
- Pixtral Large
Anthropic23
- Claude Opus 5.5
- Claude Fable 5.1
- Claude Opus 5
- Claude Opus 4.6
- Claude Fable 5
- Claude Opus 4.7
- Claude Sonnet 5.5
- Claude Opus 4.8
- Claude Sonnet 4.6
- Claude Opus 4.5
- Claude Sonnet 5
- Claude Sonnet 4.5
- Claude Opus 4.1
- Claude Haiku 5.5
- Claude Haiku 4.5
- Claude Opus 4
- Claude Sonnet 4
- Claude 3.7 Sonnet
- Claude 3.5 Sonnet
- Claude 3 Opus
- Claude 3.5 Haiku
- Claude 3 Sonnet
- Claude 3 Haiku
xAI22
Z.ai17
NVIDIA14
- NVIDIA-Nemotron-3-Ultra-550B-A55B
- NVIDIA-Nemotron-3.5-Super-VL-120B-A12B
- NVIDIA-Nemotron-3-Super-120B-A12B
- NVIDIA-Nemotron-3-Nano-30B-A3B
- Llama 3.3 Nemotron Super 49B v1.5
- NVIDIA-Nemotron-3.5-Lightning-30B-A3B
- Llama 3.1 Nemotron Ultra 253B v1
- Llama 3.3 Nemotron Super 49B v1
- Llama 3.1 Nemotron 70B Instruct
- Llama 3.1 Nemotron 51B Instruct
- Nemotron-4 340B Instruct
- Llama 2 70B SteerLM Chat
- Nemotron 3 Ultra
- nvila-internal-15b-v1
DeepSeek11
Tencent11
Ai210
Moonshot AI9
IBM9
Microsoft9
Xiaomi7
StepFun7
MiniMax7
Amazon5
Cohere5
Inception4
01.AI4
AI21 Labs4
Reka3
Fireworks AI3
Baidu2
ByteDance2
Thinky2
Reflection AI2
Upstage2
Ant Group2
Arcee2
Liquid AI2
Poolside2
Perplexity2
Meituan1
Nous Research1
Snowflake1
TII1
inclusionAI1
Synlan1
Deep Cogito1
Other labs40
- INTELLECT-3
- Athene V2 Chat
- Athene 70B
- InternLM2.5 20B Chat
- Zephyr ORPO 141B A35B v0.1
- Starling LM 7B Beta
- Tulu 2 DPO 70B
- DBRX Instruct Preview
- Starling LM 7B Alpha
- OpenChat 3.5
- Vicuna 33B
- Zephyr 7B Beta
- MPT 30B Chat
- Zephyr 7B Alpha
- Vicuna 13B
- Guanaco 33B
- SmolLM2 1.7B Instruct
- StripedHyena Nous 7B
- Vicuna 7B
- Koala 13B
- ChatGLM3 6B
- MPT 7B Chat
- RWKV-4 Raven 14B
- ChatGLM2 6B
- Alpaca 13B
- ChatGLM 6B
- OpenAssistant Pythia 12B
- FastChat-T5 3B
- StableLM Tuned Alpha 7B
- Dolly 2.0 12B
- Composer 2.5
- InternVL2 26B
- InternVL2 4B
- KAT-Coder-Pro V1
- LLaVA 1.6 34B
- LLaVA-OneVision Qwen2 72B OV
- Ling 3.0 Flash Fin
- MiniCPM-V 2.6
- Mixtral
- SWE-1.6 Fast
Hover or focus a model to see its numbers. Each tile opens its page.
Open weightsProprietaryBest for the price
Each dot is a model. Up is better, left is cheaper. The line joins the models no other model beats on both rating and price.
Leaders
LMArena Text
Rating- 1Gemini 4 Argon1526
- 2Claude Opus 5.51516
- 3Claude Fable 5.11513
- 4Claude Opus 51508
- 5Claude Opus 4.61502
Vals Index
Accuracy- 1Claude Sonnet 5.570.8%
- 2Claude Opus 5.570.2%
- 3Gemini 4 Argon69.2%
- 4Muse Wonder Full66.8%
- 5GPT-6 Astra65.0%
LMArena Coding
Rating- 1Claude Opus 5.51545
- 2Gemini 4 Argon1539
- 3Claude Opus 4.61534
- 4Claude Opus 51533
- 5MiMo-V2.6-Pro1533
SWE-bench
Accuracy- 1Claude Opus 597.0%
- 2GPT-5.6 Sol96.2%
- 3Grok 4.695.6%
- 4GPT-5.6 Terra95.4%
- 5GLM-5.395.4%
LMArena WebDev
Rating- 1Claude Opus 5.51807
- 2GPT-6 Astra1774
- 3Claude Sonnet 5.51769
- 4GPT-6.1 Sol1749
- 5Claude Fable 5.11743
Terminal-Bench 4.0
Accuracy- 1Claude Opus 5.565.2%
- 2Claude Sonnet 5.564.1%
- 3Kimi K3.1 Pro63.6%
- 4GPT-6 Astra59.6%
- 5Claude Fable 5.158.1%
Most downloaded open models
Downloads, 30 daysAll models
Compare models| # | Model | Lab | ||||||
|---|---|---|---|---|---|---|---|---|
| 1 | Gemini 4 Argon | 1526 | — | 69.2% | $2 / $10 | 1M | 2026-10-13 | |
| 2 | Claude Opus 5.5 | Anthropic | 1516 | — | 70.2% | $4 / $20 | 1M | 2026-09-22 |
| 3 | Claude Fable 5.1 | Anthropic | 1513 | — | — | $10 / $50 | 1M | 2026-09-01 |
| 4 | Claude Opus 5 | Anthropic | 1508 | — | — | $5 / $25 | 1M | 2026-07-24 |
| 5 | Claude Opus 4.6 | Anthropic | 1502 | +4 | — | $5 / $25 | 1M | 2026-02-04 |
| 6 | Muse Wonder | Meta | 1501 | — | 64.9% | $4 / $20 | 1M | 2026-10-12 |
| 7 | Gemini 3.8 Flash | 1498 | — | 56.0% | $0.82 / $4.13 | 1M | 2026-09-02 | |
| 8 | MiMo-V2.6-Pro | Xiaomi | 1491 | — | 57.8% | $0.43 / $0.87 | 1.1M | 2026-09-21 |
| 9 | Muse Spark 1.3 | Meta | 1491 | — | — | $1.25 / $4.25 | 1M | 2026-09-02 |
| 10 | Claude Fable 5 | Anthropic | 1491 | −2 | — | $10 / $50 | 1M | 2026-06-09 |
| 11 | Claude Opus 4.7 | Anthropic | 1490 | +6 | — | $5 / $25 | 1M | 2026-04-16 |
| 12 | Gemini 3.7 Flash | 1490 | — | — | $0.75 / $3.75 | 1M | 2026-08-13 | |
| 13 | Qwen3.8-Max | Alibaba | 1484 | +3 | 59.0% | $2 / $6 | 1M | 2026-09-03 |
| 14 | Gemini 3.5 Flash | 1482 | −1 | — | $1.50 / $9 | 1M | 2026-05-19 | |
| 15 | Muse Spark 1.2 | Meta | 1482 | −8 | — | $1.25 / $4.25 | 1M | 2026-08-05 |
| 16 | Gemini 3.1 Pro Preview | 1480 | +0 | — | $2 / $12 | 1M | 2026-02-19 | |
| 17 | Gemini 3.6 Flash | 1480 | +0 | — | $0.75 / $3.75 | 1M | 2026-07-21 | |
| 18 | Gemini 3 Pro | 1480 | 0 | — | — | — | 2025-11-16 | |
| 19 | Muse Spark 1.1 | Meta | 1478 | −2 | — | $1.25 / $4.25 | 1M | 2026-07-16 |
| 20 | Qwen3.7-Max-Preview | Alibaba | 1475 | +1 | — | $1.48 / $4.42 | 1M | 2026-05-21 |
| 21 | Kimi K3 | Moonshot AI | 1475 | +2 | 53.0% | $0.24 / $11 | 1M | 2026-07-16 |
| 22 | Muse Spark | Meta | 1474 | +1 | — | — | — | 2026-04-09 |
| 23 | GLM-5.3 | Z.ai | 1474 | — | 59.7% | $0.079 / $1.40 | 1M | 2026-08-18 |
| 24 | Claude Sonnet 5.5 | Anthropic | 1473 | — | 70.8% | $2 / $10 | 1M | 2026-09-28 |
| 25 | GLM-5.3-Flash | Z.ai | 1472 | +1 | 50.9% | $0.15 / $0.50 | 1M | 2026-08-26 |
| 26 | GPT-5.5 | OpenAI | 1472 | +6 | — | $5 / $30 | 1.1M | 2026-04-24 |
| 27 | Qwen3.5-Max-Preview | Alibaba | 1470 | −1 | — | — | — | 2026-03-16 |
| 28 | GLM-5.2 | Z.ai | 1470 | +6 | — | $0.034 / $7 | 1M | 2026-06-16 |
| 29 | GPT-5.4 | OpenAI | 1469 | +17 | — | $2.50 / $15 | 1.1M | 2026-03-05 |
| 30 | ERNIE 5.1 | Baidu | 1468 | −0 | — | — | — | 2026-04-29 |
| 31 | Gemini 3 Flash | 1466 | −1 | — | $0.50 / $3 | 1M | 2025-12-17 | |
| 32 | DeepSeek-V4.1-Flash-20260910 | DeepSeek | 1464 | — | 61.3% | $0.15 / $0.60 | 1M | 2026-10-17 |
| 33 | MiMo-V2.5-Pro | Xiaomi | 1464 | −1 | — | $0.30 / $0.61 | 1.1M | 2026-04-22 |
| 34 | GLM-5.1 | Z.ai | 1463 | +0 | — | $1.40 / $4.40 | 200K | 2026-04-07 |
| 35 | Claude Opus 4.8 | Anthropic | 1461 | +8 | — | $5 / $25 | 1M | 2026-05-27 |
| 36 | Claude Sonnet 4.6 | Anthropic | 1458 | −1 | — | $3 / $15 | 1M | 2026-02-17 |
| 37 | Gemini 2.5 Pro | 1458 | 0 | — | $1.25 / $10 | 1M | 2025-06-17 | |
| 38 | GPT-5.6 Sol | OpenAI | 1457 | — | — | $4 / $20 | 1.1M | 2026-07-09 |
| 39 | Kimi K2.6 | Moonshot AI | 1455 | +0 | — | $0.65 / $3.41 | 256K | 2026-04-20 |
| 40 | Qwen3.7-Plus | Alibaba | 1455 | +0 | — | $0.32 / $1.28 | 1M | 2026-06-03 |
| 41 | MiMo-V2.6-Flash | Xiaomi | 1452 | — | 54.8% | $0.14 / $0.28 | 1.1M | 2026-09-21 |
| 42 | DeepSeek-V4-Pro | DeepSeek | 1451 | +1 | 53.0% | $0.66 / $1.98 | 1M | 2026-08-12 |
| 43 | Claude Opus 4.5 | Anthropic | 1451 | +0 | — | $5 / $25 | 200K | 2025-11-24 |
| 44 | Grok 4.20 Beta | xAI | 1451 | — | — | — | — | 2026-10-20 |
| 45 | Grok 4.20 Multi-Agent Beta | xAI | 1451 | +1 | — | — | — | 2026-03-16 |
| 46 | amazon-nova-experimental-chat | Amazon | 1449 | — | — | — | — | 2026-10-20 |
| 47 | Dola Seed 2.0 Pro | ByteDance | 1448 | +1 | — | — | — | 2026-04-02 |
| 48 | GPT-5.6 Terra | OpenAI | 1448 | — | — | $2 / $12 | 1.1M | 2026-07-09 |
| 49 | Grok 4.5 | xAI | 1448 | −2 | — | $2 / $6 | 500K | 2026-07-08 |
| 50 | Qwen3.6-Max-Preview | Alibaba | 1447 | +1 | — | — | — | 2026-04-30 |
| 51 | GLM-5 | Z.ai | 1446 | −0 | — | $1 / $3.20 | 200K | 2026-02-11 |
| 52 | Kimi K2.5 | Moonshot AI | 1445 | — | — | $0.50 / $2.50 | 256K | 2026-01-27 |
| 53 | NVIDIA-Nemotron-3-Ultra-550B-A55B | NVIDIA | 1445 | — | — | — | — | 2026-10-20 |
| 54 | ERNIE 5.0 | Baidu | 1445 | +0 | — | — | — | 2025-11-16 |
| 55 | Grok 4.20 Beta 1 | xAI | 1445 | +0 | — | — | — | 2026-02-19 |
| 56 | GPT-6.1 Sol | OpenAI | 1444 | — | 63.9% | $2 / $10 | 1.1M | 2026-09-29 |
| 57 | Gemma 4 31B | 1443 | +1 | — | $0.20 / $0.40 | 256K | 2026-04-02 | |
| 58 | GPT-5.1 | OpenAI | 1443 | +20 | — | $1.25 / $10 | 400K | 2025-11-13 |
| 59 | Claude Sonnet 5 | Anthropic | 1442 | — | — | $2 / $10 | 1M | 2026-06-30 |
| 60 | GPT-6 Astra | OpenAI | 1442 | — | 65.0% | $10 / $50 | 1.1M | 2026-09-04 |
About the Explorer
How are models ranked?
Two ways. LMArena ratings come from people comparing two anonymous models side by side and voting for the better answer. Vals AI scores come from fixed tests of real work, like fixing GitHub issues, analysing filings or answering legal questions. We show both, each with its source and date.
Where do prices come from?
Each model’s list price per million input and output tokens, from its provider. Blended prices weigh input three to one against output, a common mix for chat and coding work.
How often is it updated?
Whenever LMArena or Vals AI publish new results. This snapshot is from Oct 24, 2026.
Who publishes the Explorer?
Nexus Labs. It is independent of every lab: the same layout and rules for every model, every number from a public source with its date, and corrections logged.
Can I use this data?
Yes, with attribution: download it as JSON. LMArena’s data is CC BY 4.0; cite LMArena and Vals AI for their numbers.