Meet Kimi AI by Moonshot AI - the complete AI platform for coding, research, writing, and autonomous agent workflows. Now featuring K2.7 Code HighSpeed at 260 tok/s.
Six major releases in twelve months - each one pushing a specific capability frontier while building on the same 1T-parameter MoE foundation.
The same Kimi K2.7 Code model - Moonshot's most capable coding model - served at extreme throughput. Up to 260 tokens per second on short-context tasks, ~180 tok/s on median coding inputs. Designed for agentic workflows where speed determines task completion time. Rolling out to Kimi Code Beta, API developers, and Business users.
⚡ 260 tok/s peak · 6× faster $1.90 / $8.00 per 1MMoonshot's most capable coding model to date. Reduces reasoning token usage by ~30% vs K2.6 while improving scores on every benchmark. Mandatory thinking mode, preserve-thinking across turns. MCP tool-use SOTA: 81.1 on MCP Mark Verified, beating Claude Opus 4.8 at 76.4. Multimodal via MoonViT 400M encoder - text, image, and video input. Open weights under Modified MIT license.
+21.8% Kimi Code Bench v2 −30% reasoning tokensLong-horizon agentic coding, general availability. 300-agent swarm with 4,000 coordinated steps. Claw Groups for cross-model collaboration. Document-to-Skill conversion. 262K context window. SWE-bench Verified 80.2% - SOTA among open-source models. BrowseComp Swarm 86.3%. Supports instant mode and thinking mode. The backbone of the Kimi platform for most users.
80.2% SWE-bench Verified 300 agents · 4,000 stepsNative multimodal intelligence trained on 15T mixed visual + text tokens. MoonViT 400M encoder for images and video. 256K context. Agent Swarm v1: 100 parallel sub-agents, 1,500 tool calls, 4.5× execution speedup. SWE-bench 76.8%, AIME 96.1%, VideoMMU 86.6%. The breakthrough multimodal release that established Kimi's vision capabilities.
100 sub-agents · 4.5× fasterPost-trained reasoning variant with interleaved chain-of-thought and native tool use. 200–300 sequential tool calls without losing task context. Native INT4 quantization via QAT for 2× speed vs FP16 - no accuracy loss. Tencent CodeBuddy integrates K2 Thinking as its core engine. Pioneered the think→act→observe→think loop at production scale.
300 tool calls · 2× speed INT4The original trillion-parameter open-source frontier. 1T MoE, 32B active, 128K context, MuonClip optimizer for zero training instability across 15.5T tokens. Set the open-source agentic baseline: SWE-bench 65.8%, MMLU-Pro 73.3%, τ²-bench 80%. Modified MIT License. Still a strong cost-efficient production model for text-only workflows.
Open source · Modified MITUp to 260 tokens per second on short-context tasks, ~180 tok/s on median coding inputs. 6× faster than standard K2.7 Code. Same model, optimized serving infrastructure. Now rolling out to Beta users.
K2.6 coordinates up to 300 sub-agents executing 4,000 steps simultaneously. Compress hours of parallel research and code generation into minutes. Available from Allegretto plan upward.
K2.7 Code always reasons before responding - no shortcutting on complex tasks. Preserve-thinking keeps the chain across multi-turn sessions. 30% fewer reasoning tokens than K2.6 means lower cost per session.
MoonViT 400M encoder processes images and video frames alongside text. Upload screenshots, design mockups, diagrams, or video recordings. Works in K2.5, K2.6, and K2.7 Code. Design-to-code from any visual input.
K2.7 Code scores 81.1 on MCP Mark Verified - beating Claude Opus 4.8 at 76.4. Reliable invocation of GitHub, Notion, Filesystem, Postgres, and Playwright tools in CI/CD and agent loop workflows.
Direct AI-native access to World Bank datasets, financial market data, economic indicators, and academic research - queryable in plain language inside Kimi workflows. Available from Adagio (200 requests/mo).
Migrate from GPT or Claude by changing two lines: base_url and model. Full function-calling, streaming, structured outputs, and tool use. Token-based billing separate from membership.
~192,000 words per session - entire codebases, full document stacks, long agentic sessions. K2.6 has 262K; K2.7 Code and K2.5 have 256K. K2 original has 128K. Extended via Mooncake efficient serving architecture.
All K2 models ship as open weights on HuggingFace under Modified MIT License. Deploy with vLLM, SGLang, KTransformers, or TensorRT-LLM. Commercial use permitted for most deployments. Full infrastructure control for privacy-critical workloads.
Start free with 200 Professional Data requests/month and access to Kimi K2 Adaptive mode, Kimi Slides, and Deep Research - no credit card required.
Access all models, tools, and agent features. Free to start.
Start immediately, no credit card needed. You get:
The recommended starting point for regular professional use:
Token-based API billing separate from membership. Get a key at platform.kimi.ai. K2.6: $0.55/1M input · $2.65/1M output. K2.7 Code: $0.95/1M input · $4.00/1M output. K2.7 HighSpeed: $1.90/1M · $8.00/1M.
Write, convert, review, and translate documents. Upload PDFs, Word files, slides, or spreadsheets - get clean structured outputs with summaries and professional formatting. Powered by K2.6 document agent. LaTeX PDF support.
Try Kimi Docs →Two creation modes: Adaptive (30–60 min, research-first with K2 Thinking) and Visual (5–10 min with Nano Banana Pro / Gemini 3 Pro Image). Chart parsing converts static chart images to native editable PPTX objects. Agentic Slides converts any document to a deck.
Try Kimi Slides →Build formulas, pivots, and dashboards from plain language. Process up to 1M rows of data in a single session. Generate charts, financial models, pivot tables, and data summaries. Export to Excel-compatible formats.
Try Kimi Sheets →Create modern, responsive websites from a prompt. Visual Coding with K2.6 turns design screenshots directly into production-ready code. Full-stack generation: frontend, backend, auth, and database ops in one session.
Try Website Builder →Terminal-first coding agent powered by K2.7 Code. Works in VS Code, Cursor, JetBrains, and Zed. Autonomous coding sessions, codebase navigation, multi-file editing. Starting at $19/month with 3× agent quota. K2.7 Code HighSpeed Beta available.
Try Kimi Code →24/7 cloud-based AI agent - no server setup needed. Persistent long-term memory. 5,000+ ClawHub skills. 40GB cloud storage. Scheduled automation and 24/7 task execution. Available on Allegretto plan and above.
Try Kimi Claw →Fully OpenAI and Anthropic SDK compatible. Change two lines to switch from GPT or Claude. Token-based billing, separate from app membership.
kimi-k2.7-code-highspeed - 260 tok/s ⚡
kimi-k2.7-code - Coding specialist
kimi-k2.6 - General purpose GA
kimi-k2.5 - Multimodal vision
kimi-k2-thinking - Reasoning chain
kimi-k2-instruct - Conversational
Cached input: $0.19/1M (all models). Get API key →
K2.7 Code: thinking always on · temperature=1.0
K2.6: thinking switchable (on/off)
K2.5: thinking switchable
K2 Thinking: temperature=1.0
K2 Instruct: temperature=0.6
Access chain-of-thought via reasoning_content in response. Pass it back in multi-turn sessions to preserve thinking context.
Download from HuggingFace → moonshotai. Deploy with vLLM, SGLang, KTransformers, or TensorRT-LLM. Block-fp8 format. Run the Kimi Vendor Verifier before production traffic. A100-class GPU minimum for K2-series models.
Access Kimi K2 on web and mobile - no setup, no credit card. Use Instant mode for quick answers, Thinking mode for deep reasoning, Agent mode for multi-step tasks. Free tier includes 200 Professional Data requests/month and Kimi Slides Adaptive mode.
Get your key at platform.kimi.ai. Set base_url="https://api.moonshot.ai/v1" and model="kimi-k2.7-code" in any OpenAI SDK. Works with LangChain, LlamaIndex, and Anthropic SDKs. K2.7 Code HighSpeed: kimi-k2.7-code-highspeed.
Terminal-first coding agent at kimi.com/code. Works in VS Code, Cursor, JetBrains, and Zed. Default model: K2.7 Code since June 12, 2026. Join Beta Program at kimi.com/code/beta for HighSpeed access.
Download from huggingface.co/moonshotai. Deploy with vLLM, SGLang, KTransformers, or TensorRT-LLM. Block-fp8 format. Kimi Vendor Verifier available for deployment validation. Modified MIT License - commercial use permitted.
Kimi uses music tempo-inspired plan names. All tiers include access to Kimi's full tool ecosystem. API billing is always separate from membership.
All plans support 256K context. API billing is always separate from membership. Full pricing details →
Kimi K2.7 Code Moonshot AI · Jun 2026 |
Kimi K2.6 Moonshot AI · Apr 2026 |
GPT-5.5 OpenAI |
Claude Opus 4.8 Anthropic |
DeepSeek V4 Pro DeepSeek |
Gemini 2.5 Pro Google |
|
|---|---|---|---|---|---|---|
| Architecture & Access | ||||||
| Open weights | ✓ MIT | ✓ MIT | ✗ | ✗ | ✓ | ✗ |
| Parameters (total / active) | 1T / 32B | 1T / 32B | ~200B | ~200B | 671B / 37B | ~1T est. |
| Context window | 256K | 262K | 1M (Pro) | 200K | 1M | 1M |
| Multimodal (image/video) | ✓ MoonViT | ✓ | ✓ | ✓ | ✓ | ✓ |
| Benchmarks (mix of vendor + independent) | ||||||
| SWE-bench Verified | Pending (indep.) | 80.2% ✓ | 58.6%* | 64.3%* | ~55% | ~56% |
| MCP Mark Verified | 81.1 ★ | ~73 | - | 76.4 | - | - |
| Kimi Code Bench v2 (vendor) | 62.0 | 50.9 | 69.0* | 67.4* | - | - |
| AIME 2026 | - | 96.4% | - | - | - | - |
| Speed & Cost | ||||||
| API input $/1M | $0.95 | ~$0.55 | $5.00 | $15.00 | $1.74 | $1.25 |
| API output $/1M | $4.00 | ~$2.65 | $30.00 | $75.00 | $3.48 | $5.00 |
| HighSpeed serving | 260 tok/s ⚡ | ~60 tok/s | ~80 tok/s | ~60 tok/s | ~90 tok/s | ~80 tok/s |
| −30% reasoning tokens | ✓ vs K2.6 | - | - | - | - | - |
| Platform & Ecosystem | ||||||
| Agent Swarm (parallel agents) | 300 (K2.6) | 300 | Limited | Limited | Limited | Limited |
| Integrated office tools | Docs/Slides/Sheets/Code | Full suite | Canvas | Artifacts | Limited | Workspace |
| Free tier with agent mode | ✓ 3/day | ✓ | Limited | Limited | - | Limited |
*SWE-bench Pro figures. Kimi Code Bench v2 competitor scores run by Moonshot - GPT-5.5 in Codex xhigh, Opus 4.8 in Claude Code xhigh. Throughput figures are approximate. Always verify current benchmarks and pricing on official pages.