Kimi K2.7 Code: Features, Limits, and Price
moonshotai/kimi-k2.7-codeA MoonshotAI coding model with image input, tool use, and a 262,144-token context window.
262K tokens of context — about 196,608 words, on a scale that ends at 2M.
What it can do
- Reasoning
- Vision
- Tool use
- Open weights
- Takes text
- Takes image
- Gives text
Released 12 Jun 2026 · checked 3 Oct 2026.
About it
Kimi K2.7 Code is MoonshotAI’s coding-focused model for completing programming tasks across long inputs. It accepts text and images, can use tools, and produces text answers. The model has open weights and a 262,144-token context window. Usage costs $0.6712 per million input tokens and $3.35 per million output tokens.
Review
In a line
Kimi K2.7 Code is MoonshotAI’s model for coding work. It belongs to the Kimi K2 family and is available as an open-weight model.
What it's good at
This model is aimed at end-to-end programming tasks, including work that needs a long amount of information at once. It supports reasoning and tool use.
It can read text and images, then return text. Its open weights and listing as a MoonshotAI model may suit people comparing open-weight models.
Price in plain terms
Input costs $0.6712 per million tokens, while output costs $3.35 per million tokens. Input tokens are the text or images sent in; output tokens are the words the model generates.
Using the rough examples provided, reading a 500-word page costs about 0.044 cents. A chat turn with about 1,000 input tokens and 500 output tokens costs about 0.23 cents. A thousand such turns would cost about $2.35.
Context and limits
The context window is 262,144 tokens, so one request can include a large amount of text or other input. The maximum output is 235,929 tokens in one answer.
It returns text only, even though it can accept images.
Ways to use it
Use it in software or chat tools that support this model for programming tasks, text prompts, and image-based input. It can also call tools when the surrounding tool supports that function.
Its model identifier is moonshotai/kimi-k2.7-code, and its Hugging Face identifier is moonshotai/Kimi-K2.7-Code.
Verdict
Choose Kimi K2.7 Code if you need a coding model that can handle very large inputs, reason through tasks, work with images, and use tools. Its open weights may also matter if you want that model format.
Its output price is much higher than its input price, so long generated answers can raise the bill.
Verdict
Kimi K2.7 Code fits programmers and software builders who need one model to process lengthy project material, inspect images, and complete multi-step coding work. The large context and open weights are useful when those details matter. It is less suitable for buyers seeking image generation or a model with a lower output rate. Budgeting should focus on generated text: output tokens cost $3.35 per million, compared with $0.6712 for input.
- Handles a 262,144-token context window
- Accepts text and image inputs
- Supports reasoning and tool use
- Open weights are available
- Designed for coding tasks
- Produces text only
- Output tokens cost more than input tokens
- Maximum output is below the context limit
- Pick it ifLong programming tasks; Tool-using coding workflows; Users comparing open-weight models
- Skip it ifYou need image generation; You expect output and input to cost the same
Pricing
Input is $0.6712 per million tokens and output is $3.35 per million tokens. A typical 1,000-token-in, 500-token-out turn is about 0.23 cents; 1,000 such turns are about $2.35.
Questions
Who makes Kimi K2.7 Code?
MoonshotAI makes it. The model is part of the Kimi K2 family.
Is Kimi K2.7 Code open source?
Its weights are open. The listed Hugging Face identifier is moonshotai/Kimi-K2.7-Code.
What can it read?
It accepts text and images as input.
What does it produce?
It produces text output.
How much context can it handle?
Its context window is 262,144 tokens.
What does it cost?
Input costs $0.6712 per million tokens. Output costs $3.35 per million tokens.
How much does a typical chat turn cost?
A turn with about 1,000 input tokens and 500 output tokens costs about 0.23 cents. A thousand such turns cost about $2.35.
Does it support tools and reasoning?
Yes. The model supports tool use and reasoning.