Moonshot: Kimi K2.6 API
Kimi K2.6 is a native multimodal, trillion-parameter mixture-of-experts (MoE) model engineered for state-of-the-art agentic reasoning and long-horizon task execution. Building on the foundation of K2.5, it expands the "Agent Swarm" architecture to support up to 300 parallel sub-agents coordinating over 4,000 tool-call steps. It features a dual-mode system: "Instant Mode" for rapid dialogue and "Thinking Mode" for deep reasoning, capable of autonomous runs lasting over 12 hours. In 2026, Kimi K2.6 is recognized as the premier open-weight alternative for complex software engineering, multi-day deep research, and large-scale document orchestration.
- Context window: 256,144 tokens
- Max output: 16,384 tokens
- Input: text, video, file, image
- Output: text
- Reasoning: Supported
- Tool calling: Supported
- File input: Supported
- Released: 2026-04-19
- Knowledge cutoff: 2025-11-30
Frequently Asked Questions
What is the context window of Kimi K2.6?
Kimi K2.6 supports a context window of up to 256,144 tokens.
Does Kimi K2.6 support function calling?
Yes. Kimi K2.6 supports tool / function calling.
Does Kimi K2.6 support reasoning?
Yes. Kimi K2.6 is a reasoning-capable model.
What is the knowledge cutoff of Kimi K2.6?
The knowledge cutoff of Kimi K2.6 is 2025-11-30.