Cost to run

Qwen3-235B-A22B-Thinking-2507

Qwen/Qwen3-235B-A22B-Thinking-2507

Family: Qwen3
Context: 262,144 tokens

Qwen3-235B-A22B-Thinking-2507 can be run self-hosted (rent a GPU + run vLLM/TGI) or through a serverless API (pay per token). Live pricing comparisons:

Self-hosted

Rent a GPU from Lambda, RunPod, Vast.ai, CoreWeave. Full control.

Serverless

Pay per token via Together, Fireworks, DeepInfra, Groq.

Need to benchmark this model against another? Try the calculator or see where it ranks on the InferenceScore leaderboard.