Kimi K2

Earlier Kimi open-weight MoE model still useful as a baseline for agentic coding and tool-use comparisons.

Moonshot AI · Kimi

Editorial review

Reviewed byOpenSourcesAI EditorialLast updatedAugust 2026SourcesMoonshot AI Kimi K2

Model checkpoints, context windows, provider support, local runtime compatibility, and license terms can change quickly. Verify the exact model card before production or commercial use.

Best for

Builders comparing earlier Kimi releases against newer Kimi K2.6 behavior.

Who should use it

Builders comparing earlier Kimi releases against newer Kimi K2.6 behavior.
Teams with access to hosted inference or server-class deployment paths.
Developers evaluating coding assistant, repo-editing, and code review workflows.
Teams testing tool-use, agentic planning, and multi-step workflow behavior.

Common workflows

Agentic coding, tool use, long-horizon tasks
agents workflows
coding workflows
moe workflows
legacy workflows

Deployment and hardware notes

Server-class for full weights; use hosted providers for practical evaluation.

License and usage notes

Modified MIT. Open weights. Verify the exact model card and license terms for the checkpoint or hosted provider you use.

Strengths

Open weights model option for Kimi workflows.
Builders comparing earlier Kimi releases against newer Kimi K2.6 behavior.
Tracked as Legacy baseline in the OpenSourcesAI model directory.

Limitations

Not the main current Kimi entry; compare against the latest Kimi release before standardizing.
Server-class for full weights; use hosted providers for practical evaluation.
Context window and limits: 131,072 tokens (RoPE-extended).
Verify the exact model card, provider docs, license, and serving support before production use.