Kimi K2.6
LLM Provider / Model · Long-horizon coding, UI/UX generation from prompts, multi-agent orchestration, cost-optimized frontier-class workloads (~80% cheaper than GPT-5.5)
At a glance
| Input price | $0.60 |
| Output price | $2.50 |
| Cache price | $0.06 |
| Price tier | Budget |
| Context window | 256K |
| Max output | 32K |
| Context tier | 128K-500K |
| Speed tier | Standard |
| Latency | medium |
| Knowledge cutoff | Jan 2026 |
| Modality | Multimodal (vision) |
| Model ID | kimi-k2.6 |
| Provider | MoonshotAI |
| HumanEval | 92% |
| MMLU | 87% |
| SWE-Bench | 59% |
| Benchmark | SWE-bench Pro: top (ties GPT-5.5) |
| Released | Apr 2026 |
| Hosting | Open-weights |
| Capabilities | Vision, Tool use, Streaming, Structured output, Agentic coding, Multi-Head Latent Attention, INT4 quantization, 1T params (32B active), Multi-agent orchestration, Self-hostable |
What Kimi K2.6 does
Vision, Tool use, Streaming, Structured output, Agentic coding, Multi-Head Latent Attention, INT4 quantization, 1T params (32B active), Multi-agent orchestration, Self-hostable
Best for
Long-horizon coding, UI/UX generation from prompts, multi-agent orchestration, cost-optimized frontier-class workloads (~80% cheaper than GPT-5.5)
Works well with
Integration
Agent / Orchestration
Conflicts & caveats
- Trae's built-in model menu may not include Kimi K2.6 — Trae primarily ships with Claude/GPT/Gemini/DeepSeek.
- Google Antigravity is built around Gemini 3 Pro; Claude and recent GPT-5 models also work, but Kimi K2.6 is not in the supported model list.
- Kiro's spec mode is tuned for Claude models — Kimi K2.6 may work but spec-driven flows are best with Claude.
- OpenAI Codex (VS Code) is tied to OpenAI's GPT-5.1 Codex family — Kimi K2.6 is not selectable inside this integration.
- Gemini CLI is locked to Google Gemini models — your LLM is Kimi K2.6.
- Privacy conflict: Self-hosted Kimi K2.6 sends code to cloud @codebase (Cursor). Use local context (Continue indexing, ChromaDB, LanceDB, pgvector, Vespa self-hosted) for true privacy.
- Claude Code only works with Claude models — your LLM is Kimi K2.6.
- Devin uses Cognition's proprietary model internally — your selected Kimi K2.6 will be ignored.
- OpenAI Codex (Cloud) runs GPT-5.1 Codex internally — your selected Kimi K2.6 will be ignored.
- Jules runs Gemini models internally — your selected Kimi K2.6 will be ignored.
- The Claude Code SDK is tuned for Claude models — Kimi K2.6 may work via the model adapter but loses Claude-specific features (thinking, prompt caching, sub-agents).
- Lovable uses a fixed internal model — your selected Kimi K2.6 will be ignored.
- Replit Agent 3 uses its own model selection internally — your selected Kimi K2.6 will be ignored.
Build a full stack around Kimi K2.6 — Flowpicker shows compatibility warnings before you commit.
Open the stack planner →