Moonshot AI 开源 Kimi K2.7-Code：编程模型在 Kimi Code Bench v2 上较 K2.6 提升 21.8%

K2.7-Code is a Mixture-of-Experts model. It holds 1T total parameters and activates 32B per token. The design uses 384 experts, with 8 selected per token and 1 shared. It has 61 layers, including 1 dense layer.

Attention uses MLA, and the feed-forward path uses SwiGLU. A MoonViT vision encoder adds 400M parameters for image and video input. The model ships with native INT4 quantization. The context window is 256K tokens (262,144).

Two constraints matters: Thinking mode is mandatory; disabling it returns an API error. Sampling is fixed: temperature 1.0, top_p 0.95, n 1, penalties 0.0. Default max output is 32,768 tokens.

You can self-host with vLLM, SGLang, or KTransformers. The Hugging Face repository is large, roughly 595 GB on disk. This is a server-class deployment target, not a laptop model.

Benchmark

Moonshot team published six benchmark rows. They compare K2.7-Code against K2.6, GPT-5.5, and Claude Opus 4.8. K2.7-Code beats K2.6 on every row. The largest coding jump is Kimi Code Bench v2, from 50.9 to 62.0.

K2.7-Code does beat Opus 4.8 on MCP Mark Verified, 81.1 versus 76.4. It also lands close to GPT-5.5 on MLS Bench Lite. K2.7-Code ran in Kimi Code CLI, GPT-5.5 in Codex xhigh, and Opus 4.8 in Claude Code xhigh.

Reasoning-Token Efficiency: A Cost Claim, Not Just Quality

Moonshot team reports about 30% lower reasoning-token usage than K2.6. It frames this as ‘less overthinking.’

Reasoning tokens bill as output tokens on most price cards. Agentic coding runs hundreds or thousands of steps. Each plan, retry, and verification pays the thinking cost again. A 30% cut compounds across a long run.

The effect lands in three places at once. First, lower output-token cost per task. Second, faster steps, which helps interactive CLI sessions. Third, more steps before hitting context limits.

Use Cases With Examples

• Repo-scale refactors are the main use case. Point the agent at a failing test suite. It reads files, edits across modules, then reruns tests until green.
• Code review is a second fit. Feed a pull request diff and ask for risk analysis. The 256K window holds large diffs, logs, and related files together.
• MCP tool-use workflows are a third fit. K2.7-Code scored 81.1 on MCP Mark Verified. That suite tests correct tool invocation through the Model Context Protocol. Think CI checks, ticket updates, and file edits in one loop.
• Long-context analysis is a fourth fit. The model accepts text, image, and video input. Documentation, screenshots, and a recorded repro can share one prompt.

Marktechpost’s Interactive Explorer

Kimi K2.7-Code — Interactive Explorer

Company-reported benchmarks and official API pricing. Released June 12, 2026. Verified June 12, 2026.

Benchmarks

Cost Calculator

Specs

Source: Moonshot AI Kimi K2.7-Code model card. K2.7-Code ran in Kimi Code CLI; GPT-5.5 in Codex xhigh; Claude Opus 4.8 in Claude Code xhigh. First-party numbers, not an independent leaderboard.

Input cost

Output cost

Est. monthly total

Rates: cached input $0.19 / 1M, cache-miss input $0.95 / 1M, output $4.00 / 1M (official Kimi pricing). Savings line illustrates K2.7-Code’s reported ~30% lower reasoning-token usage vs K2.6, applied to the reasoning share of output. Estimate only.

Source: Kimi K2.7-Code Hugging Face model card and Kimi API docs.

A Minimal Quickstart

The Kimi API is OpenAI-compatible. The model string is kimi-k2.7-code. Do not override the fixed sampling parameters, or the request errors.

import os
from openai import OpenAI

# Base URL and key per the Kimi API docs at platform.moonshot.ai
client = OpenAI(
    api_key=os.environ.get("MOONSHOT_API_KEY"),
    base_url="https://api.moonshot.ai/v1",
)

messages = [
    {"role": "system", "content": "You are a coding agent."},
    {"role": "user", "content": "Refactor utils.py to remove duplicate code."},
]

resp = client.chat.completions.create(
    model="kimi-k2.7-code",
    messages=messages,
    max_tokens=32768,  # default cap; also the maximum
    # thinking is enabled by default and cannot be disabled.
    # temperature (1.0), top_p (0.95), n (1), and penalties (0.0) are
    # fixed server-side. Passing any other value returns an error.
)

msg = resp.choices[0].message
print(msg.content)

# Multi-step tool calls: append the full assistant message so that
# reasoning_content is preserved. Dropping it errors on the next turn.
# messages.append(msg.model_dump())

Two tool-use rules come from the docs. Keep reasoning_content from the current turn in context. And set tool_choice to only "auto" or "none".

How K2.7-Code Compares

K2.7-Code lists $0.19 per 1M for cached input.

Strengths and Weaknesses

Strengths:

• Open weights under Modified MIT, with a real self-host path.
• Broad, consistent gains over K2.6 on coding and agent evals.
• Low API pricing relative to closed frontier models.
• Beats Opus 4.8 on the MCP Mark Verified benchmark (company-reported).

Weaknesses:

• All headline numbers are first-party at launch.
• Thinking mode cannot be disabled.
• Sampling controls are locked to fixed values.
• Multi-step tool calls must preserve reasoning_content.
• 595 GB weights make self-hosting a serious commitment.

Key Takeaways

• All headline benchmarks are vendor-run; independent results are pending.
• K2.7-Code is open-weight, coding-specialized, and built on Kimi K2.6.
• Moonshot reports +21.8% on Kimi Code Bench v2 over K2.6.
• The model uses roughly 30% fewer reasoning tokens than K2.6.