Spiral gives your agent every tool with one account. Browse the tools →

Codex · LLMReleased 2026-07-16

How to Use Kimi K3 in Codex

Kimi K3 suits long coding tasks and research that turns into working reports or dashboards. Its million-token context supports large projects, but explicit boundaries help prevent unrequested changes.

Prompts to try with Kimi K3 in Codex

Copy an example into Codex, or ask in your own words.

Use Kimi K3 to trace the failing tests in this repository to their root cause. Make the smallest justified fix, run the relevant tests, and report the evidence. Do not refactor unrelated code.

Kimi K3Codex
Try asking
  • Use Kimi K3 to implement the requirements in SPEC.md. First identify affected modules and acceptance checks, then make the changes and run those checks. Stop and report any conflicting requirements.

  • Use Kimi K3 to profile this project's slowest GPU kernel. Compare each revision against the current implementation using the same inputs and numerical tolerance. Keep only changes supported by measured results.

  • Use Kimi K3 to turn the research method in docs/method.md into a Python implementation. Map equations to functions, document assumptions, and validate the results against the examples in that document.

  • Use Kimi K3 to build an interactive report from this project's research data. Include source references beside quantitative claims, test the calculations, and list missing data instead of inventing values.

How to set up Kimi K3 in Codex

Pick the way that suits you. Each takes about two minutes.

Use Spiral's MCP server

  1. Add the hosted server
    codex mcp add spiral --url https://api.spiral.ai/mcp
  2. Sign in

    Codex opens the browser: sign in to Spiral and click Authorize. No API key. codex mcp list then shows Spiral.

    codex mcp login spiral

Use the Spiral skill

  1. Send Codex this message

    Your agent reads Spiral's skill and sets itself up.

    set up https://spiral.ai/SKILL.md
  2. Sign in or paste a key

    Approve Spiral in the browser, or give the agent a key from the console.

Use the Spiral CLI

  1. Install and sign in

    Installs the Spiral CLI, signs you in and adds its skill to your coding agents.

    npx -y spiral-cli@latest init

Call the Spiral API

  1. Create an API key

    In the console, under API keys.

  2. Run Kimi K3

    One request; the result comes back by polling or a webhook.

    curl https://api.spiral.ai/v1/runs \
      -H "Authorization: Bearer $SPIRAL_API_KEY" \
      -H "Content-Type: application/json" \
      -H "Idempotency-Key: $(uuidgen)" \
      -d '{
        "provider": "openrouter",
        "model": "moonshotai/kimi-k3",
        "input": {
          "messages": [
            {
              "role": "system"
            }
          ]
        }
      }'

Then paste one of the prompts above: Codex finds Kimi K3 on Spiral, shows the price and runs it.

Kimi K3 providers and prices

Spiral runs Kimi K3 through OpenRouter. Open a card for today's price, with every fee included.

OpenRouterLLM
MoonshotAI: Kimi K3
moonshotai/kimi-k3

Settings you can ask for

Ask Codex for any of these in your prompt.

SettingAllowed valuesDefaultDescription
Seeda whole number
Min p0 to 1
Top ka whole number from 0
Logprobson or off
Max tokens1 to 943718

The most output tokens to generate.

Top logprobs0 to 20
Reasoning effortnone, low, high, or maxmax

How much the model reasons.

Repetition penalty0 to 2
Parallel tool callson or off

Every API field of MoonshotAI: Kimi K3 is on its tool page: open its card above.

What can you do with Kimi K3 in Codex?

Turn a specification into tested code across a large project with Kimi K3. Set the permitted changes and acceptance checks so a long coding task stays focused.

Fix bugs across a large codebase

Use Kimi K3 to trace a failing behavior through several modules, propose a focused fix, and check it against a reproducible test.

Implement a feature from a specification

Give Kimi K3 the requirements, permitted changes, and acceptance checks. Ask it to carry the implementation through testing and report any unfinished requirements.

Optimize a GPU kernel

Ask Kimi K3 to profile a kernel, investigate the bottleneck, and compare revisions against the same correctness tolerance and benchmark.

Reproduce a computational research method

Use Kimi K3 to connect a paper's equations to executable code, document assumptions, and compare computed results with the published method.

Build an interactive research report

Ask Kimi K3 to organize supplied evidence into a report with charts and drill-down views. Require a source for each quantitative claim.

Draft a research presentation

Turn a research brief into a slide narrative with chart specifications, supporting evidence, and speaker notes using Kimi K3.

Create a recurring analysis dashboard

Use Kimi K3 to turn fresh data into dashboard updates, explanations of changes, and a list of gaps that need investigation.

Questions about Kimi K3 in Codex

Can Codex use Kimi K3?

Yes. Add Spiral to Codex as an MCP server, and Codex can find and run Kimi K3 like any built-in tool.

How much does Kimi K3 cost in Codex?

Up to $0.085206 per run at the default inputs. $0.2496 per 1M input tokens, $20.80 per 1M output tokens. You pay from your Spiral balance as you go, with no subscription, and a failed run is not charged. Your Codex plan is separate.

Do I need an OpenRouter account to use Kimi K3 in Codex?

No. Spiral runs Kimi K3 on its own OpenRouter account. You sign in to Spiral once and never paste a key.

How long does it take to set up Kimi K3 in Codex?

About two minutes: add the hosted server, then sign in.

Does Codex wait while Kimi K3 runs?

Kimi K3 runs in the background. Codex starts the run, checks on it and gives you the text, json when it's ready.

What is Kimi K3 good at?

Kimi K3 is suited to coding tasks that span many steps, large codebases, and research that becomes a report, presentation, or dashboard. Give it a concrete deliverable and a way to check the result.

How much context can Kimi K3 handle?

Moonshot documents a context window of one million tokens. Treat that as room for relevant evidence and conversation history, rather than a target to fill with unrelated material.

Which reasoning setting should I choose?

Moonshot documents low, high, and maximum effort, with maximum as the default. Start with low for bounded tasks and compare higher effort on work that requires several dependent decisions.

What should I watch for with Kimi K3?

Moonshot warns that ambiguous instructions can lead to unexpected decisions. State which changes are allowed, what needs review, and when the task is finished.

Why does conversation history matter for Kimi K3?

Kimi K3 expects complete assistant messages to be retained across conversation turns and tool calls, including reasoning and tool-call information. Keeping only the visible answer discards information Moonshot says the model requires.

Get started

Start using Kimi K3 in Codex

One Spiral account for every tool your agent needs. No subscription; you pay per run.

Paste this into Codex:

Set up Codex

set up https://spiral.ai/SKILL.md
Are you an AI agent? See setup options