Moonshot AI Kimi K2.6 Now Available on Workers AI
Key Points
- Kimi K2.6 multimodal agentic model now available on Workers AI
- 262.1k token context window with competitive benchmark performance
- Agent swarm orchestration supporting 300 sub-agents and 4,000 coordinated steps
Summary
@cf/moonshotai/kimi-k2.6 is now available on Workers AI in partnership with Moonshot AI. Kimi K2.6 is a native multimodal agentic model built on a Mixture-of-Experts architecture with 1T total parameters and 32B active per token, delivering frontier-scale intelligence with efficient inference.
Key Points
- Context & Performance: 262.1k token context window; scores competitively against GPT-5.4 and Claude Opus 4.6 on agentic and coding benchmarks (BrowseComp: 83.2, SWE-Bench Verified: 80.2, Terminal-Bench 2.0: 66.7)
- Core Capabilities: Long-horizon coding across Rust, Go, and Python; coding-driven design for production-ready interfaces; agent swarm orchestration supporting 300 sub-agents with 4,000 coordinated steps; vision inputs and multi-turn tool calling
- API Changes from K2.5: Reasoning controlled via
chat_template_kwargs.thinking(replacesenable_thinking); reasoning content returned inreasoningfield (replacesreasoning_content) - Access Methods: Available through Workers AI binding (
env.AI.run()), REST API (/ai/run), OpenAI-compatible endpoint (/v1/chat/completions), and AI Gateway