# Kimi K2.7 Code

> **TL;DR.** Kimi K2.

- **Category:** AI / Frontier Models / Open-Weight LLMs
- **Stage:** nascent
- **Age:** 2 days
- **Origin date:** 2026-06-12
- **First detected:** 2026-06-13
- **Canonical URL:** https://earlyterms.com/term/kimi-k2-7-code
- **Sources:** 7 primary URLs

## Definition

Kimi K2.7 Code is an open-weight, coding-focused agentic LLM from Moonshot AI built on a Mixture-of-Experts architecture: 1 trillion total parameters, 32 billion active per token, 256K token context window, and mandatory reasoning enabled on every inference call.

Moonshot released weights on [Hugging Face](https://huggingface.co/moonshotai/Kimi-K2.7-Code) on June 12, 2026 under a Modified MIT license, priced at $0.95 per million input tokens via API. The model targets long-horizon software engineering tasks — repository-scale refactors, multi-step agentic workflows, and multi-language codebases — claiming 30% fewer reasoning tokens than K2.6.

## Analogy

Think of it as a self-driving engineer that thinks less but codes more by the mile.

## Why it's emerging now

Moonshot AI dropped Kimi K2.7 Code on June 12, 2026 with open weights, a 30% reasoning-token reduction over K2.6, and pricing up to 12x below Claude Fable 5 on output tokens — making it the most cost-efficient frontier-adjacent coding agent available for self-hosting teams as of this week.

## Related terms

- *parent:* Kimi K2.6
- *competitor:* DeepSeek V4 Pro
- *competitor:* Qwen3
- *competitor:* GLM-4.7
- *competitor:* Claude Opus 4.8
- *competitor:* GPT-5.5
- *parent:* agentic coding
- *related:* coding agents
- *related:* Model Context Protocol
- *child:* Kimi Code
- *competitor:* MiniMax M3

## Sources

1. [Hugging Face — moonshotai/Kimi-K2.7-Code (model card + weights)](https://huggingface.co/moonshotai/Kimi-K2.7-Code)
2. [Kimi official resource page](https://www.kimi.com/resources/kimi-k2-7-code)
3. [Kimi API quickstart documentation](https://platform.kimi.ai/docs/guide/kimi-k2-7-code-quickstart)
4. [The Decoder — K2.7 Code undercuts GPT-5.5 and Claude by up to 12x](https://the-decoder.com/moonshots-open-model-kimi-k2-7-code-undercuts-gpt-5-5-and-claude-by-up-to-12x-on-price-per-token/)
5. [MarkTechPost — Moonshot AI releases Kimi K2.7-Code, +21.8% on Kimi Code Bench v2](https://www.marktechpost.com/2026/06/12/moonshot-ai-releases-kimi-k2-7-code-a-coding-model-reporting-21-8-on-kimi-code-bench-v2-over-k2-6/)
6. [CryptoBriefing — Kimi AI releases open-source K2.7 Code model with 1T parameters](https://cryptobriefing.com/kimi-k2-7-code-open-source-release/)
7. [Hacker News — flagship launch thread (445 points, 235 comments)](https://news.ycombinator.com/item?id=48502347)

---
_Generated by EarlyTerms · https://earlyterms.com/term/kimi-k2-7-code_