Moonshot AI · in Borg

Kimi K2.7 Code

Kimi K2.7 Code is Moonshot AI’s coding-focused successor to K2.6, retaining its vision-capable architecture while improving long-horizon software engineering and instruction following.

Reasoning262,144 tokenstext, image, video

262,144 tokensContext window

4780 msMedian time to first token (gateway proxy)

24.2Output tokens per second (whole turn)

$0.95 / $4Vendor API price in / out · per 1M

Included in Borg

Download Borg Borg for Business

At a glance

Specification	Value
Provider	Moonshot AI
Family	Kimi
Context window	262,144 tokens Vendor source
Input modalities	text, image, video
Reasoning	Yes
Open weights	Yes
Borg lane ids	`kimi-for-coding`, `kimi-for-coding-highspeed`, `moonshotai/kimi-k2.7-code`
Median time to first token (gateway proxy) on Borg	4780 ms
Median tokens per second (end to end) on Borg	24.2 output tokens per second

Not published by the vendor: release date, output modalities, thinking modes, knowledge cutoff, fusion roles.

What Kimi K2.7 Code is good for

Kimi K2.7 Code is Moonshot AI’s coding-focused successor to K2.6, retaining its vision-capable architecture while improving long-horizon software engineering and instruction following. Unlike the general-purpose K3 flagship, this smaller K2-series model concentrates on repository work, tool use and sustained implementation at a substantially lower direct API price. Thinking is always enabled, and prior reasoning is preserved across turns, which suits coding agents that need to keep track of decisions through multiple edits and tests. It accepts images and, through the official API, video, so an agent can use screenshots or clips alongside code. The high-speed API variant is the same model with faster output, not a separate entry here. Borg includes K2.7 Code in the subscription; Borg Fusion can route coding tasks to it, or users can select its lane directly.

multi-file coding with iterative test runs
long-context instruction-following in codebases
tool-using development agents
visual inspection during frontend fixes

How Borg uses it

Kimi K2.7 Code is available as a subscription catalog lane in Borg. The catalog does not publish which Fusion profiles route to this model, so no routing role is claimed here. You can pick it directly by setting the OpenAI-compatible gateway model field to a listed lane id. Check the current catalog for availability before building a workflow around a lane.

model: "kimi-for-coding"

Measured on Borg

Metric	Median	95th percentile	Samples
Time to first token (gateway proxy)	4780 ms	13025 ms	119
Tokens per second (end to end)	24.2 output tokens per second	61.8 output tokens per second	1894

Time to first token (gateway proxy)

Lower is fasterFleet median 3971.5 ms

Kimi K2.7 Code4780 ms

Output tokens per second (whole turn)

Higher is fasterFleet median 31.9 tokens/s

Kimi K2.7 Code24.2 tokens/s

Time to first token (gateway proxy): time until the first streamed response event reaches the gateway; our closest available measure of time to first token
Tokens per second (end to end): tokens per second, end to end: provider-reported output tokens divided by the whole turn's elapsed time, including queueing and reasoning; not a streaming per-token rate

Completion speed covers the whole turn including queueing and reasoning time; it is not a streaming tokens-per-second figure.

Measured on Borg: SearchAtlas AI Gateway production request logs. 30-day window ending 2026-10-08; 2583 requests. Minimum 100 samples per published request metric.

Benchmarks

Benchmark	Vendor-reported score	Source
Kimi Code Bench v2	62	Vendor source
Program Bench	53.6	Vendor source
MLS Bench Lite	35.1	Vendor source
Kimi Claw 24/7 Bench	46.9	Vendor source
MCP Atlas	76	Vendor source
MCP Mark Verified	81.1	Vendor source

Pricing

Included in Borg subscription. Vendor API list price: $0.95 input / $4 output per 1M tokens. Vendor price source.

Compare subscription and API pricing · Cost Savings Calculator

FAQ

What is Kimi K2.7 Code?

Kimi K2.7 Code is listed in Borg's subscription catalog as kimi-for-coding. Kimi K2.7 Code is Moonshot AI’s coding-focused successor to K2.6, retaining its vision-capable architecture while improving long-horizon software engineering and instruction following.

Is Kimi K2.7 Code in Borg?

Yes. Kimi K2.7 Code is listed in Borg's subscription catalog under the lane kimi-for-coding. You can choose the lane directly, or explore the available Fusion profiles. Availability and model details can change; check the current catalog before deployment.

What context window does Kimi K2.7 Code have?

The published context window is 262,144 tokens. Consult the cited sources for details and current limits. Vendor source

Is Kimi K2.7 Code good for coding?

Vendor-published evaluations appear in the benchmarks section; their settings may differ from your tasks. Its listed uses include multi-file coding with iterative test runs, long-context instruction-following in codebases, tool-using development agents, visual inspection during frontend fixes. Try the lane for your own tasks, and compare the output with your requirements before choosing a default.

How fast is Kimi K2.7 Code on Borg?