Kimi

Excellent

by Moonshot AI · founded 2023 · updated Jul 2026

Kimi is Moonshot AI's assistant and open-weight model family (the K2 line), built as a trillion-parameter mixture-of-experts that activates a fraction of its weights per token. It punches at the frontier on agentic tool use and coding while pricing far below Western closed labs, and the open weights let teams self-host or fine-tune for full data control.

LLMs & AssistantsOpen SourceEstablishedFrom Free chat / from $0.60 per 1M tokens (API)

Visit website ↗Compare

84.0TIP Score

Excellent

The score, taken apart

Fixed weights, sources attached, the formula

Capability(30%)88

Value for Money(15%)93

Security & Compliance(15%)68

Integrations & Ecosystem(15%)80

Maturity & Reliability(10%)78

Momentum(15%)91

Best for

◆Cost-sensitive frontier LLM workloads
◆Self-hosted agentic coding
◆Teams fine-tuning open models

Key integrations

OpenAI-compatible APIOpenRouterOpen weights (Hugging Face)MCP / agentic tool usevLLM

Strengths

＋Frontier-class agentic coding — K2.7-Code leads Opus 4.8 on MCP-Mark Verified
＋Trillion-parameter MoE with open weights for self-hosting and fine-tuning
＋256K context and API pricing that undercuts most closed competitors many times over

Limitations

－China-based hosting raises data-residency and compliance questions for regulated buyers
－Trails the very top on some pure-reasoning benchmarks (GPQA-Diamond, AIME) versus GPT-5.4
－English-language ecosystem, docs and enterprise support thinner than US labs

Evidence behind this score

Every TIP Score is backed by verifiable claims. Data is a curated snapshot, always confirm current terms with the vendor.

Claim	Source	As of
Kimi K2.6 priced at $0.60 input / $2.50 output per 1M tokens on the official API	OpenRouter / Moonshot pricing	Jun 2026
K2.7-Code (1T total / 32B active MoE, 256K context) beats Opus 4.8 on MCP-Mark Verified, 81.1 vs 76.4	OpenRouter model card & MarkTechPost	Jun 2026
K2.6 trails GPT-5.4 on GPQA-Diamond (90.5% vs 92.8%) and AIME 2026 (96.4% vs 99.2%)	llm-stats.com benchmarks	Jun 2026

Intelligence on Kimi

Adoptionmedium impactJul 11, 2026

Kimi K2.7 Code becomes first open-weight model in Copilot's picker

GitHub added Moonshot's Kimi K2.7 Code to Copilot's model picker — the first open-weight model offered there. Copilot's multi-model strategy now spans OpenAI, Anthropic, Google and open weights, giving teams a cheaper agentic-coding option inside the tool they already use.

GitHub Copilot KimiGitHub changelog

Releasemedium impactJul 10, 2026

Moonshot's Kimi K2.7-Code beats Opus 4.8 on agentic coding at open-weight prices

Moonshot AI's K2 line — a trillion-parameter mixture-of-experts with 256K context and open weights — is rattling the coding-model market: K2.7-Code leads Opus 4.8 on MCP-Mark Verified (81.1 vs 76.4) while the API prices from $0.60 per 1M tokens. A strong option for cost-sensitive and self-hosted agentic coding, with data-residency caveats for regulated buyers.

Kimi DeepSeek Claude CodeOpenRouter, MarkTechPost & Moonshot pricing