siGit Code + Kimi K3
Kimi K3 is currently selectable with siGit Code on AgentSky. siGit Code controls the work loop and tools; the model supplies reasoning and generation. Start a real task below with both choices already selected.
How the combination works
siGit Code runs the work. Kimi K3 powers each turn.
Runs the task through its ACP harness in a persistent cloud workspace while the selected AgentSky model supplies reasoning and generation.
AgentSky supplies the persistent cloud computer, task state, usage records, and enabled tools or connectors around that combination.
siGit Code
Owns the agent loop, repository or computer operations, tool calls, and how results are returned.
Kimi K3
Moonshot AI's model supplies the reasoning and generated output within siGit Code's supported interface.
AgentSky
Keeps the selected pair, workspace, task state, access configuration, and itemized usage together.
Current Kimi K3 facts
What the model is
Moonshot's flagship multimodal model, with always-on reasoning and a selectable effort level.
Context window
1M tokens in the current AgentSky catalog.
Accepted input
Text, Image, Video in the current AgentSky catalog.
Model token price
$3 / $15 per 1M input / output tokens on the displayed AgentSky plan. Runtime and paid tool usage are separate.
Evaluate it on real work
Start with one bounded task and evidence you can inspect.
Choose one of the platform-verified models, provide a bounded task and add only the project or tools the work needs.
Open the preselected Composer
The Composer above starts with siGit Code and Kimi K3. Add the repository or project and only the tools the task needs.
Adapt this first task
Inspect the workspace, complete one bounded change or analysis, and return the result with evidence that can be checked.
Verify the outcome
Inspect the actual output or diff and reproduce the important check; a successful ACP session alone does not establish task correctness.
Fit and limits
When to choose siGit Code + Kimi K3
Tasks supported by this ACP harness through AgentSky's verified platform-model path.
This page establishes
- The pair is selectable in the current AgentSky compatibility matrix
- The launch carries both selections into the shared Composer
- Current catalog facts are shown when the gateway publishes them
You still need to measure
- Quality on your own repository or workflow
- End-to-end task time, recovery effort, and human review
- Total task cost, which also depends on runtime, tools, connectors, and token usage
Common questions
Related resources