Kimi K3.1 Pro launch and setup
Learn about Kimi K3.1 Pro for coding and reasoning, its BetterToken model ID, Claude Code setup, and supported APIs.
Updated: October 24, 2026
Kimi K3.1 Pro is a Moonshot AI reasoning model for coding, research, and multi-step agent tasks. You can evaluate it for feature development, debugging, and source analysis.
BetterToken supports Model ID kimi-k3.1-pro. Check the model plaza for current availability and pricing.
Model features
Kimi specifies a 1-million-token context window. Thinking is enabled by default, with low, high, and max levels in the official API. Context limits and configurable thinking parameters depend on your client and API.
Quick setup
- Create a BetterToken API Key, or use your existing Key.
- Set Model ID to
kimi-k3.1-pro, preserving case and punctuation. - Choose your protocol, Base URL, and authentication field from the table.
| Usage | Base URL | Authentication field | Request path |
|---|---|---|---|
| OpenAI-compatible clients | https://bettertoken.ai/v1 | Authorization: Bearer YOUR_BETTERTOKEN_API_KEY | POST /v1/chat/completions |
| Claude Code / Anthropic Messages | https://bettertoken.ai | x-api-key: YOUR_BETTERTOKEN_API_KEY | POST /v1/messages |
In Claude Code configured for BetterToken, enter /model kimi-k3.1-pro to switch models. For initial configuration, see the Claude Code setup guide.
Responses is not enabled for this model on BetterToken. Do not use it with client configurations that require the Responses API.
Streaming and thinking output
Both APIs support streaming and non-streaming requests. With Chat Completions, read reasoning_content and content separately. Preserve the complete assistant message in multi-turn conversations. See the Chat Completions API.