# kimi-k2.8-preview — API, Pricing & Context Window | Vivgrid

> Source: https://vivgrid.com/docs/models/kimi-k2.8-preview

kimi-k2.8-preview on Vivgrid: Moonshot AI's efficient coding model with a 1M-token context window and tool calling, at a fraction of kimi-k3's price.

`kimi-k2.8-preview` is Moonshot AI's efficient coding model, released in September 2026. It brings the Kimi family's **1M-token context window** to a mid-tier price point — **\$1.00 per 1M input tokens**, **\$0.26 cached**, and **\$4.15 output** — roughly a third of what `kimi-k3` costs, while keeping repository-scale context.

Its defining feature is **adjustable thinking effort**. Dial reasoning down for routine edits and up for hard problems, so you pay for deep reasoning only when a task needs it.

Vivgrid serves `kimi-k2.8-preview` through the unified, **OpenAI-compatible Chat Completions API** (`/chat/completions`), so you can route Moonshot alongside OpenAI, Anthropic, and Google models with one API key and one bill.

## Specifications

| Field | Value |
| --- | --- |
| Provider | Moonshot AI |
| Model ID | `kimi-k2.8-preview` |
| Best for | Coding |
| Available on | Vibe coding |
| Context window | 1,000,000 tokens |
| Modalities | Text, Image |
| Tool / function calling | Yes |
| Knowledge cutoff | — |
| Acceleration | Global (centralized) |

## Pricing

Pricing in USD per 1M tokens.

| Input | Cached input | Output |
| --- | --- | --- |
| $1.00 | $0.26 | $4.15 |

### What one request costs

The rates above applied to a 20K-token prompt with a 2K-token reply — a typical single agent turn, small enough that every request is billed at the base rate.

| Request | Cost |
| --- | --- |
| Cold prompt, nothing cached | $0.0283 |
| Warm prompt, 90% of the input served from cache | $0.0150 |


## Use kimi-k2.8-preview in your coding agent

`kimi-k2.8-preview` works with the same Vivgrid key and `https://api.vivgrid.com/v1` endpoint as every other model in the catalog. Grab your API key from the [Vivgrid Console](https://console.vivgrid.com), then drop one of the configurations below into your agent of choice.

### OpenCode

[OpenCode](https://opencode.ai) has Vivgrid built in, so you can use `kimi-k2.8-preview` without writing any config: open `/models`, search for `vivgrid`, paste your API key and pick the model. The [OpenCode tutorial](/docs/tutorials/opencode) walks through it step by step.

Prefer to configure it by hand? Install OpenCode:

```bash
npm i -g opencode-ai
```

Declare `kimi-k2.8-preview` under the Vivgrid provider in your global OpenCode config, `~/.config/opencode/opencode.json`. The `@ai-sdk/openai-compatible` provider talks to Vivgrid over **Chat Completions** (`/v1/chat/completions`):

```json title="~/.config/opencode/opencode.json" lines icon="json" highlight={3,6,9,10}
{
  "$schema": "https://opencode.ai/config.json",
  "model": "vivgrid/kimi-k2.8-preview",
  "provider": {
    "vivgrid": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Vivgrid",
      "options": {
        "baseURL": "https://api.vivgrid.com/v1",
        "apiKey": "{env:VIVGRID_API_KEY}"
      },
      "models": {
        "kimi-k2.8-preview": {
          "name": "kimi-k2.8-preview",
          "tool_call": true,
          "attachment": true,
          "modalities": { "input": ["text", "image"], "output": ["text"] }
        }
      }
    }
  }
}
```

Export your key and launch OpenCode:

```bash
export VIVGRID_API_KEY="viv-xxxxxxxxxxxxx"
opencode
```

Use `/models` to switch between `kimi-k2.8-preview` and other Vivgrid models mid-session. See the full [OpenCode tutorial](/docs/tutorials/opencode).

### Pi

Install [Pi](https://pi.dev):

```bash
npm install -g --ignore-scripts @earendil-works/pi-coding-agent
```

Add `kimi-k2.8-preview` to the Vivgrid provider in `~/.pi/agent/models.json`. Pi talks to Vivgrid over **Chat Completions** (`/v1/chat/completions`):

```json title="~/.pi/agent/models.json" lines icon="json" highlight={4,5,12}
{
  "providers": {
    "vivgrid": {
      "baseUrl": "https://api.vivgrid.com/v1",
      "apiKey": "viv-xxxxxxxxxxxxx",
      "api": "openai-completions",
      "compat": {
        "sendSessionAffinityHeaders": true
      },
      "models": [
        {
          "id": "kimi-k2.8-preview",
          "input": ["text", "image"],
          "contextWindow": 1000000
        }
      ]
    }
  }
}
```

Make it the model Pi starts with in `~/.pi/agent/settings.json`:

```json title="~/.pi/agent/settings.json" lines icon="json" highlight={2,3}
{
  "defaultProvider": "vivgrid",
  "defaultModel": "kimi-k2.8-preview"
}
```

Then verify and start Pi:

```bash
pi --list-models vivgrid
pi
```

Or run a one-shot prompt without entering the TUI:

```bash
pi -p "explain the auth flow in this repo" --model vivgrid/kimi-k2.8-preview
```

See the full [Pi tutorial](/docs/tutorials/pi).

## API surfaces

The Vivgrid Console catalogs `kimi-k2.8-preview` on one surface, all behind the same API key. The endpoint shown is the one the Console previews for that surface.

| Surface | Primary endpoint | Use it for |
| --- | --- | --- |
| Vibe coding | `/chat/completions` | Coding CLIs and IDE agents that drive the model themselves — point the tool at Vivgrid and keep your own loop. |

## Quick start

Get an API key from the Vivgrid Console (https://console.vivgrid.com), then call `kimi-k2.8-preview` directly.

```bash
curl https://api.vivgrid.com/v1/chat/completions \
  -H "Authorization: Bearer $VIVGRID_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k2.8-preview",
    "messages": [
      { "role": "user", "content": "Say hello in English, Chinese and Spanish." }
    ],
    "stream": true
  }'
```

## Choosing within the Kimi family

Prices are USD per 1M tokens at the base rate.

| Model | Context | Max output | Input | Output |
| --- | --- | --- | --- | --- |
| [`kimi-k3`](/docs/models/kimi-k3) | 1,000,000 | 128,000 | $3.00 | $15.00 |
| `kimi-k2.8-preview` (this page) | 1,000,000 | — | $1.00 | $4.15 |


## Ideal use cases

- High-volume coding agents where `kimi-k3` is more model than the task needs
- Long-context refactoring and repository-scale reading on a budget
- Tool-calling workflows that mix quick edits with occasional deep reasoning
- Cost-sensitive fallbacks behind a frontier model

## Related models

- [kimi-k3](/docs/models/kimi-k3) — Moonshot's frontier coding model
- [kimi-k2.6](/docs/models/kimi-k2.6) — the prior Kimi generation, with a 256K max output
- [minimax-m3](/docs/models/minimax-m3) — a lower-cost long-context alternative