diff --git a/docs/docs/hardware-platforms/tpu.mdx b/docs/docs/hardware-platforms/tpu.mdx index b3d7f2516..f386ebf9c 100644 --- a/docs/docs/hardware-platforms/tpu.mdx +++ b/docs/docs/hardware-platforms/tpu.mdx @@ -165,7 +165,7 @@ SGLang-JAX provides comprehensive TPU-optimized features for production LLM serv ## Optimized Model List -The following models have been tested and optimized for TPU deployment: +SGLang-JAX supports the following model families. Use the linked TPU cookbooks for deployment instructions. @@ -175,45 +175,53 @@ The following models have been tested and optimized for TPU deployment: - + - + - + - - - - - - - - - - + + - + + + + + + + + + - + - + - - + + + + + + + + + +
Model FamilyPerformance StatusCookbook
Qwen 3⭐ Recommended for productionQwen3
Qwen 3 MoE⭐ Best performanceQwen3 MoE
Qwen 2Needs improvement
Qwen 2 MoENeeds improvement
Qwen 1.5Needs improvementQwen2.5-VLQwen2.5-VL
Llama/LLaMANeeds improvementLlama 3.1 / Llama 3.3 70B
DeepSeek V3 / R1DeepSeek V3 / DeepSeek R1
Kimi-LinearKimi-Linear
Grok-2Needs improvementGrok-2
Gemma 2Verified on TPUGemma 2
Bailing MoENeeds improvementMiMo-V2-FlashMiMo-V2-Flash
MiMo-V2.5-ProMiMo-V2.5-Pro
Ling-2.6Ling-2.6