# Kimi K2.7 Code High Speed

The same Kimi K2.7 Code capabilities on a faster serving tier. You keep the coding quality and spend less time waiting on output; check the live throughput metrics on this page.

- **Model ID:** `moonshotai/kimi-k2.7-code-highspeed`
- **Type:** chat
- **Providers:** moonshotai
- **Context window:** 262,144
- **Maximum output tokens:** 32,768
- **Pricing:** $1.90/1M input tokens, $8/1M output tokens
- **Canonical page:** https://vercel.com/ai-gateway/models/kimi-k2.7-code-highspeed

## Supported parameters

Detailed capability metadata has not been reported for this model.

## Example

```ts
import { streamText } from 'ai'

const result = streamText({
  model: 'moonshotai/kimi-k2.7-code-highspeed',
  prompt: 'Why is the sky blue?'
})
```

## About

Kimi K2.7 Code High Speed is the high-speed serving tier for Kimi K2.7 Code, Moonshot AI's coding-focused agentic model. The weights and capabilities are the same; the difference is how fast the endpoint streams tokens back. Released on June 15, 2026, Kimi K2.7 Code High Speed targets coding-agent scenarios where output speed shapes the experience.

Everything documented for the standard variant carries over. Kimi K2.7 Code High Speed is a Mixture-of-Experts model with one trillion total parameters and 32 billion active per forward pass. Compared with Kimi K2.6, Kimi K2.7 Code High Speed improves coding and agent performance, spends roughly 30% fewer thinking tokens, and follows instructions better across long-horizon sessions. Native image and video input is included, and thinking mode stays always on.

Serving speed counts when a human is in the loop. Interactive coding assistants stream code while a developer watches, and edit-run-fix cycles repeat dozens of times per session. Faster output compounds across every one of those turns. Real-world speed varies with load, context length, and routing, so check the live throughput and latency metrics on this page instead of relying on published peaks.

Switching from the standard variant is a model-string change. Set the model to `moonshotai/kimi-k2.7-code-highspeed` and keep the rest of your integration: the AI SDK, Chat Completions, Responses, Messages, and other API formats all work through AI Gateway, which routes across Moonshot AI with automatic failover.

Kimi K2.7 Code High Speed supports a context window of 262.1K tokens and completions up to 32.8K tokens per request. Pricing through AI Gateway is $1.90 per million input tokens and $8 per million output tokens, with cached input at $0.38 per million tokens.

## What to consider

Kimi K2.7 Code High Speed serves the same underlying model as Kimi K2.7 Code, so output quality doesn't change; you're paying for faster serving. Compare $1.90 per million input tokens and $8 per million output tokens against the standard variant's rates before committing high-volume traffic. Thinking mode is always on, so reasoning tokens still count toward output. Throughput varies with load and context length; the live metrics on this page show measured numbers rather than published peaks.

## When to use

### Best For

- Interactive coding assistants where a developer watches output stream and slow generations stall the session
- Tight edit-run-fix loops where output speed compounds across dozens of agent turns
- Long-horizon coding agents that need Kimi K2.7 Code quality under a response-time budget
- Teams already on Kimi K2.7 Code that want faster serving without changing model behavior

### Consider Alternatives When

- Cost per token dominates. Kimi K2.7 Code serves the same model at standard rates when nobody is waiting on the stream
- The workload runs offline or in batch. Faster serving adds little without a human in the loop
- The work is general-purpose rather than code. Kimi K2.6 covers broader multimodal workflows
- Thinking mode must be optional. A non-thinking Kimi K2 variant gives you direct generation

## Best for

- **Interactive coding assistants:** Streaming code to a developer who's watching, where slow generations stall the session
- **Tight iteration loops:** Edit-run-fix cycles that repeat dozens of times, so output speed compounds across turns
- **Latency-sensitive coding agents:** Long-horizon agents that need Kimi K2.7 Code quality under a response-time budget
- **Drop-in speed upgrade:** Teams on Kimi K2.7 Code that want faster serving with identical model behavior

## Consider alternatives

- **Cost-sensitive workloads:** Kimi K2.7 Code serves the same weights at standard rates when throughput isn't the constraint
- **Batch or offline pipelines:** Faster serving adds little value when no one is waiting on the stream
- **Work beyond code:** Kimi K2.6 covers broader design-with-code and general multimodal workflows
- **Optional thinking mode:** A non-thinking Kimi K2 variant skips always-on reasoning when deliberation doesn't help

## Frequently asked questions

### How is Kimi K2.7 Code High Speed different from Kimi K2.7 Code?

Serving speed and price. The underlying model is identical, so coding quality, multimodal input, and always-on thinking all carry over. Kimi K2.7 Code High Speed streams output faster and is priced accordingly; compare the live metrics and rates on each model's page.

### How fast is Kimi K2.7 Code High Speed?

Speed varies with load, context length, and routing, so a single number would mislead. Check the live throughput and time-to-first-token metrics on this page for measured performance through AI Gateway.

### Does Kimi K2.7 Code High Speed keep the multimodal and long-horizon capabilities?

Yes. Native image and video input, the long-horizon tool-use improvements, and the leaner reasoning budget all match Kimi K2.7 Code. Only the serving tier changes.

### How do I switch from `kimi-k2.7-code` to kimi-k2.7-code-highspeed?

Update the model string in your API call to `moonshotai/kimi-k2.7-code-highspeed`. Authentication, tool-calling format, and the rest of the integration stay the same.

### How do I use Kimi K2.7 Code High Speed on AI Gateway?

Use the identifier `moonshotai/kimi-k2.7-code-highspeed` with the AI SDK or any supported interface like Chat Completions, Responses, or Messages. AI Gateway routes across `moonshotai` and handles failover automatically.

### Does Kimi K2.7 Code High Speed support zero data retention?

Yes, Zero Data Retention is available for this model. Zero Data Retention is offered on a per-provider basis. See https://vercel.com/docs/ai-gateway/capabilities/zdr for details.

## Links

- [Model page](https://vercel.com/ai-gateway/models/kimi-k2.7-code-highspeed)
- [AI Gateway documentation](https://vercel.com/docs/ai-gateway)
- [Provider model documentation](https://platform.kimi.ai/docs/pricing/chat-k27-code)