Skip to main content

​
Available models

​
By quality (LMArena Score)

ModelProviderParameterLMArena ScoreLatency per 10,000 tokens
Claude Opus 4.6Anthropicclaude-opus-4-614987.4s
Claude Opus 4.7Anthropicclaude-opus-4-71491TBD
Gemini 3.5 FlashGooglegemini-3.5-flash1480TBD
GPT-5.5OpenAIgpt-5.51475TBD
Gemini 3 Flash PreviewGooglegemini-3-flash-preview14744.2s
Claude Opus 4.5Anthropicclaude-opus-4-5-2025110114683.9s
Claude Sonnet 4.6Anthropicclaude-sonnet-4-614667.2s
Claude 4.5 SonnetAnthropicclaude-sonnet-4-5-2025092914535.6s
Gemini 2.5 ProGooglegemini-2.5-pro14484.0s
GPT-5.1OpenAIgpt-5.114392.7s
Gemini 3.1 Flash Lite PreviewGooglegemini-3.1-flash-lite-preview1438TBD
GPT-5.2OpenAIgpt-5.214371.6s
GPT-5OpenAIgpt-514344.3s
Kimi K2.5Moonshot AIkimi-k2.514321.2s
GPT-4.1OpenAIgpt-4.114131.8s
Claude 4 OpusAnthropicclaude-opus-4-20250514141213.6s
Gemini 2.5 FlashGooglegemini-2.5-flash14112.6s
Claude 4.5 HaikuAnthropicclaude-haiku-4-5-2025100114094.1s
Qwen3 Next 80B A3BAlibaba Cloudqwen3-next-80b-a3b14023.1s
GPT-5 miniOpenAIgpt-5-mini13903.8s
Claude 4 SonnetAnthropicclaude-sonnet-4-2025051413895.1s
Gemini 2.5 Flash-LiteGooglegemini-2.5-flash-lite13801.1s
gpt-oss-120bOpenAIgpt-oss-120b13531.4s
Qwen3 32BAlibaba Cloudqwen3-32B13473.7s
GPT-5 nanoOpenAIgpt-5-nano13373.2s
gpt-oss-20bOpenAIgpt-oss-20b13171.1s

​
By latency (per 10,000 tokens)

ModelProviderParameterLatency per 10,000 tokensLMArena Score
Gemini 2.5 Flash-LiteGooglegemini-2.5-flash-lite1.1s1380
gpt-oss-20bOpenAIgpt-oss-20b1.1s1317
Kimi K2.5Moonshot AIkimi-k2.51.2s1432
gpt-oss-120bOpenAIgpt-oss-120b1.4s1353
GPT-5.2OpenAIgpt-5.21.6s1437
GPT-4.1OpenAIgpt-4.11.8s1413
Gemini 2.5 FlashGooglegemini-2.5-flash2.6s1411
GPT-5.1OpenAIgpt-5.12.7s1439
Qwen3 Next 80B A3BAlibaba Cloudqwen3-next-80b-a3b3.1s1402
GPT-5 nanoOpenAIgpt-5-nano3.2s1337
Qwen3 32BAlibaba Cloudqwen3-32B3.7s1347
GPT-5 miniOpenAIgpt-5-mini3.8s1390
Claude Opus 4.5Anthropicclaude-opus-4-5-202511013.9s1468
Gemini 2.5 ProGooglegemini-2.5-pro4.0s1448
Claude 4.5 HaikuAnthropicclaude-haiku-4-5-202510014.1s1409
Gemini 3 Flash PreviewGooglegemini-3-flash-preview4.2s1474
GPT-5OpenAIgpt-54.3s1434
Claude 4.5 SonnetAnthropicclaude-sonnet-4-5-202509295.6s1453
Claude Sonnet 4.6Anthropicclaude-sonnet-4-67.2s1466
Claude Opus 4.6Anthropicclaude-opus-4-67.4s1498
Claude 4 OpusAnthropicclaude-opus-4-2025051413.6s1412
Claude Opus 4.7Anthropicclaude-opus-4-7TBD1491
GPT-5.5OpenAIgpt-5.5TBD1475
Gemini 3.1 Flash Lite PreviewGooglegemini-3.1-flash-lite-previewTBD1438
Gemini 3.5 FlashGooglegemini-3.5-flashTBD1480

​
By provider

​
Anthropic Claude

ModelParameterLMArena ScoreLatency per 10,000 tokens
Claude Opus 4.7claude-opus-4-71491TBD
Claude Opus 4.6claude-opus-4-614987.4s
Claude Sonnet 4.6claude-sonnet-4-614667.2s
Claude Opus 4.5claude-opus-4-5-2025110114683.9s
Claude 4.5 Sonnetclaude-sonnet-4-5-2025092914535.6s
Claude 4.5 Haikuclaude-haiku-4-5-2025100114094.1s
Claude 4 Opusclaude-opus-4-20250514141213.6s

​
OpenAI GPT

ModelParameterLMArena ScoreLatency per 10,000 tokens
GPT-5.5gpt-5.51475TBD
GPT-5.2gpt-5.214371.6s
GPT-5.1gpt-5.114392.7s
GPT-5gpt-514344.3s
GPT-5 nanogpt-5-nano13373.2s
GPT-5 minigpt-5-mini13903.8s
GPT-4.1gpt-4.114131.8s
gpt-oss-120bgpt-oss-120b13531.4s
gpt-oss-20bgpt-oss-20b13171.1s

​
Google Gemini

ModelParameterLMArena ScoreLatency per 10,000 tokens
Gemini 3.5 Flashgemini-3.5-flash1480TBD
Gemini 3 Flash Previewgemini-3-flash-preview14744.2s
Gemini 3.1 Flash Lite Previewgemini-3.1-flash-lite-preview1438TBD
Gemini 2.5 Progemini-2.5-pro14484.0s
Gemini 2.5 Flashgemini-2.5-flash14112.6s
Gemini 2.5 Flash-Litegemini-2.5-flash-lite13801.1s
Gemini 3.1 Flash Lite Preview is currently available in the US region only.

​
Alibaba Cloud Qwen

ModelParameterLMArena ScoreLatency per 10,000 tokens
Qwen3 Next 80B A3Bqwen3-next-80b-a3b14023.1s
Qwen3 32Bqwen3-32B13473.7s

​
Moonshot AI Kimi

ModelParameterLMArena ScoreLatency per 10,000 tokens
Kimi K2.5kimi-k2.514321.2s
Claude Opus 4.5 and Claude Opus 4.6 currently support context windows under 200k tokens via the LLM Gateway.
For information on data retention and model training policies for each provider, see Data Retention and Model Training.
Head to our Playground to test out LLM Gateway without having to write any code!