Skip to main content
Geo-specific inference keeps inference processing for supported serverless models inside one geography, with no GPUs to provision. You send requests to that geography’s API endpoint instead of api.together.ai, using the same API key and model IDs.

Choose a model

These models are available through the US endpoint:
  • moonshotai/Kimi-K3.
  • zai-org/GLM-5.3.
  • zai-org/GLM-5.3-Flash.
  • deepseek-ai/DeepSeek-V4.1-Flash.
  • deepseek-ai/DeepSeek-V4.1-Flash-1017.
  • deepseek-ai/DeepSeek-V4-Pro-0813.
  • MiniMaxAI/MiniMax-M3.1-Flash.

Send a request

Set your client’s base URL to https://us-api.together.ai/v1 for US processing. The examples use GLM-5.3.
Requests to us-api.together.ai keep inference processing in the United States. Requests to api.together.ai do not guarantee US processing.

Pricing

US-only inference costs 1.25× the standard serverless rate (a 25% premium) for input, cached input, and output tokens. See US-only prices for per-model rates.

Limitations

  • Inference processing only: The guarantee covers where your request is processed. It doesn’t cover where data is stored, where logs are written, or how network traffic is routed.
  • No fallback outside the selected geography: If a model isn’t available in the selected geography, the request fails with a 404 error of type region_unavailable instead of running somewhere else. If your workload requires processing in a specific geography, send retries to that same endpoint, and don’t fall back to api.together.ai.
For private networking or other data-residency requirements, see privacy and security.