Skip to main content
These models use the Anthropic Messages-compatible request and response format. They are separate model families; protocol compatibility does not make them Claude models. For Claude models, see the Claude gateway.

Compare Messages models and pricing

Prices are in USD per 1 million tokens. Prices verified: 2026-11-18. Savings compare standard input prices with regular market reference prices. GLM standard prices are 75% of the reference price, Kimi prices are 80%, and DeepSeek and Grok prices are 60%. Account-specific prices may differ. A $0 cache-write rate means no separate cache-write charge in this rate card. See API Pricing.

What to change

Quickstart

Change only model to use another ID from this page. deepseek-flash is the published call ID for DeepSeek V4.1 Flash. Use Messages for the Grok text models on this page; the Grok image and video task APIs are separate products.

Streaming and usage

Add "stream": true to the same request body to receive Messages SSE events. Responses report actual token usage, including cache reads and writes when present. Billing separates ordinary input, output, cache-read and cache-write tokens. Keep the API key server-side and use API Pricing for current rates.

Supported Messages models

  • glm-5.3
  • glm-5.2
  • glm-5.1
  • deepseek-flash
  • deepseek-v4-pro
  • kimi-k3.1-pro
  • kimi-k3
  • kimi-k2.7-code
  • kimi-k2.6
  • grok-4.7
  • grok-4.6
  • grok-4.5

Remote MCP availability

Use list_models with integration_type: llm_gateway to discover these model IDs. Pass the selected ID to get_model_schema for the Messages contract or to get_integration_guide for server-side SDK setup. APIXO MCP