Kimi K2.7 Code — Especificações
Compiled by Mustafa Ihsan from the vendor’s published documentation · Last updated
| Desenvolvedor | Moonshot AI |
|---|---|
| Tipo | LLM (codificação, MoE) |
| Modalidade | Texto → Texto |
| Parâmetros | 1T no total / 32B ativos (MoE) |
| Janela de contexto | 256K |
| Licença | MIT modificada (aberta) |
| Pesos abertos | Sim |
| Lançado | 2026-06 |
| Preço da entrada | $0.6 /1M |
| Preço da saída | $2.5 /1M |
| Provedores de API | Moonshot, OpenRouter |
Execute-o localmente
| VRAM (4 bits) | ~500 GB |
|---|---|
| GPU mínima | Servidor multi-GPU |
What is Kimi K2.7 Code?
Kimi K2.7 Code is Moonshot AI’s open coding model — a 1-trillion-parameter
mixture-of-experts with 32B active parameters, 384 experts across 61 layers, tuned
specifically for agentic software engineering. It has a 256K context, ships under a
modified-MIT licence, and uses roughly 30% fewer reasoning tokens than its predecessor.
That last figure is the one that pays. In an agentic coding loop the model reasons on
every step, and reasoning tokens are billed as output — so a 30% reduction in tokens spent
thinking is close to a 30% cut in the running cost of the agent, independent of the headline
price. At $0.60 in / $2.50 out per million tokens, K2.7 Code is already an order of magnitude
below frontier coding models; the token efficiency compounds that. The 256K context is
narrower than the 1M offered by several rivals, which matters for whole-repository reasoning
but rarely for the file-and-dependency scope most coding agents actually operate on.
Self-hosting needs around 500 GB at 4-bit, so this is an API model in practice for all but
the largest deployments.
Kimi K2.7 Code pricing: API cost per 1M tokens
| Entrada (por 1 milhão de tokens) | $0.600 |
|---|---|
| Saída (por 1 milhão de tokens) | $2.50 |
| Razão saída/entrada | 4.2× |
| Misturada (4:1 entrada:saída) | $0.980 por 1 milhão de tokens |
What Kimi K2.7 Code costs per month
Gasto mensal real com uma proporção de entrada para saída de 4:1 — a razão efetivamente gerada por cargas de trabalho típicas de chat ou RAG.
| Carga de trabalho | Tokens/mês | Custo por mês |
|---|---|---|
| Projeto paralelo | 1 milhão de tokens de entrada / 0,25 milhão de tokens de saída | $1.23 |
| Equipe pequena | 20 milhões de tokens de entrada / 5 milhões de tokens de saída | $25 |
| Produção | 200 milhões de tokens de entrada / 50 milhões de tokens de saída | $245 |
Calcule seus próprios números na Calculadora de custos de API de IA.
Cheaper alternatives to Kimi K2.7 Code
| Modelo | Custo médio ponderado por US$ 1 milhão | Você economiza |
|---|---|---|
| DeepSeek V4-Pro aberta | $0.522 | 47% cheaper |
| DeepSeek V4-Flash aberta | $0.168 | 83% cheaper |
Hospedar localmente ou pagar pela API?
Kimi K2.7 Code is open-weight, so you can run it yourself. It needs ~500 GB of VRAM at 4-bit (Multi-GPU server). Self-hosting only beats the API once your volume is high enough to keep that hardware busy — the calculadora de autohospedagem versus API calcula o ponto de equilíbrio para seu volume de tokens.
Perguntas frequentes
How much does Kimi K2.7 Code cost per 1M tokens?
Kimi K2.7 Code costs $0.600 per 1M input tokens and $2.50 per 1M output tokens. At a typical 4:1 input-to-output mix that blends to about $0.980 per 1M tokens.
How much does Kimi K2.7 Code cost per month?
A small-team workload of 20M input and 5M output tokens a month costs about $25 on Kimi K2.7 Code. A side project (1M in / 0.25M out) costs roughly $1.23.
What is a cheaper alternative to Kimi K2.7 Code?
DeepSeek V4-Pro is the strongest cheaper option in our database at $0.522 per 1M blended — about 47% less than Kimi K2.7 Code. It is also open-weight, so self-hosting is an option.
Can I run Kimi K2.7 Code locally?
Yes. Kimi K2.7 Code is open-weight and needs about ~500 GB of VRAM at 4-bit quantisation (Multi-GPU server).
Why does Kimi K2.7 Code charge more for output than input?
Output tokens are generated one at a time and cannot be batched the way a prompt can, so they cost the provider more to serve. Kimi K2.7 Code charges 4.2× more for output, which is why prompt-heavy workloads are far cheaper to run than generation-heavy ones.
Os preços correspondem às tarifas listadas oficialmente para a API principal do modelo e são revisados conforme os provedores os atualizam. Descontos por volume, processamento em lote ou entradas em cache não estão incluídos. Compare todos os modelos lado a lado na Banco de dados de modelos de IA ou o Ranking de LLMs.