ContextQMD
Back to Promptfoo

Moonshot (Kimi)

site/docs/providers/moonshot.md

0.121.186.0 KB
Original Source

Moonshot (Kimi)

Moonshot AI provides an OpenAI-compatible API for its Kimi models — the Kimi K2 thinking models and the Moonshot v1 generation models. The Moonshot provider extends the OpenAI provider, so all of its options are supported.

Setup

  1. Get an API key from the Kimi (Moonshot) platform.
  2. Set the MOONSHOT_API_KEY environment variable or specify apiKey in your config.
yaml
providers:
  - id: moonshot:kimi-k2.6

Both moonshot:<model> and moonshot:chat:<model> resolve to the chat completions endpoint. If you omit the model, the provider defaults to kimi-k2.6.

Available Models

Moonshot's lineup rotates over time — call the list models API (GET https://api.moonshot.ai/v1/models) for the live set. As of writing:

  • Kimi K2 — thinking models, 256k context: kimi-k2.6, kimi-k2.5, kimi-k2.7-code, kimi-k2.7-code-highspeed. These reason before answering and emit a separate reasoning stream (see below).
  • Moonshot v1 — generation models: moonshot-v1-8k, moonshot-v1-32k, moonshot-v1-128k (context-length variants), the vision variants moonshot-v1-8k-vision-preview / moonshot-v1-32k-vision-preview / moonshot-v1-128k-vision-preview, and the auto-router moonshot-v1-auto.

The older kimi-k2-0711-preview, kimi-k2-0905-preview, kimi-k2-turbo-preview, kimi-k2-thinking, kimi-k2-thinking-turbo, and kimi-latest ids were discontinued in 2026 — use kimi-k2.6 instead.

Configuration

yaml
providers:
  - id: moonshot:kimi-k2.6 # flagship thinking model — leave sampling params unset
  - id: moonshot:moonshot-v1-8k # generation model — accepts arbitrary sampling
    config:
      temperature: 0.2
      max_tokens: 1024

Configuration Options

The provider accepts every option the OpenAI provider supports. Commonly used:

  • temperature, max_tokens, top_p, presence_penalty, frequency_penalty
  • stream
  • response_format (JSON mode), tools / tool_choice (function calling)
  • showThinking — set to false to drop a thinking model's reasoning from the graded output (default true)
  • cost, inputCost, outputCost, cacheReadCost — Moonshot ships no built-in price table, so set these to track cost. Every override is in USD per token. Moonshot's official pricing page publishes rates in USD per 1 million tokens, so divide each published rate by 1,000,000 before configuring it. inputCost/outputCost take precedence over the flat cost; cacheReadCost prices cached prompt tokens.

For example, these illustrative per-million rates convert to per-token overrides. Check the official pricing page for the current rates for your model before using them.

yaml
providers:
  - id: moonshot:kimi-k2.6
    config:
      inputCost: 0.00000095 # $0.95 per 1M tokens divided by 1,000,000
      cacheReadCost: 0.00000016 # $0.16 per 1M tokens divided by 1,000,000
      outputCost: 0.000004 # $4.00 per 1M tokens divided by 1,000,000

Any other parameter supported by the OpenAI provider is forwarded as-is.

Kimi K2 thinking models

The kimi-k2.x models are reasoning models and behave differently from the moonshot-v1 family:

  • Fixed sampling parameters. Kimi pins temperature (1.0 with thinking on), top_p, n, and the penalties to fixed values and returns a 400 ("invalid temperature: only 1 is allowed for this model") for any other value. The provider therefore does not send promptfoo's default temperature/max_tokens for kimi-* models — leave them unset (recommended) or set temperature: 1. The moonshot-v1 models accept arbitrary sampling values.
  • Reasoning output. Kimi returns a separate reasoning_content stream that promptfoo surfaces with a Thinking: … prefix. Set showThinking: false when you assert on structured output (for example is-json) so the reasoning doesn't contaminate the parsed result.
  • Token budget. Reasoning tokens count against max_tokens. When you leave max_tokens unset the provider lets Moonshot apply its 32k default; if you set it, leave generous headroom for the answer.
  • Disable thinking. kimi-k2.6 and kimi-k2.5 support thinking: { type: disabled } (pass it via config.passthrough); kimi-k2.7-code is always thinking.
yaml
providers:
  - id: moonshot:kimi-k2.6
    config:
      showThinking: false
      passthrough:
        thinking: { type: disabled } # optional: turn reasoning off

See Using Thinking Models for the full behavior matrix.

Vision

The vision models (moonshot-v1-*-vision-preview) and the multimodal Kimi models (kimi-k2.5 / kimi-k2.6 / kimi-k2.7-code) accept base64-encoded image input using the standard OpenAI image_url content format. Moonshot does not accept remote image URLs — embed images as data: URIs. See Use the Kimi Vision Model.

Example Usage

yaml
providers:
  - id: moonshot:kimi-k2.6
  - id: openai:gpt-4o-mini

prompts:
  - 'Summarize the following in one sentence: {{text}}'

tests:
  - vars:
      text: 'Promptfoo is an open-source tool for testing and evaluating LLM apps.'

A runnable comparison lives in examples/provider-moonshot.

API Details

  • Base URL: https://api.moonshot.ai/v1 (global). China-mainland keys use https://api.moonshot.cn/v1 — point at it with apiBaseUrl, since the global and China platforms issue region-locked keys.
  • OpenAI-compatible chat completions API.
  • Full API documentation.

See Also