Use CoralBricks with Kilo Code

Run Kilo Code's agent on CoralBricks open-model inference — GLM and Kimi with up to 1M context — as a custom provider with auto-detected models.

Kilo Code is an open-source AI coding agent for VS Code, JetBrains, and the terminal. CoralBricks connects as a custom provider — and Kilo auto-detects our models, context windows and all, from the gateway's /v1/models endpoint.

1. Get an API key

Create a key from your API keys page (format ak_...).

2. VS Code / JetBrains

Settings (gear icon) → Providers → scroll down → Custom provider:

Provider ID — coralbricks; Display name — CoralBricks
Provider API — OpenAI Compatible
Base URL — https://inference.coralbricks.ai/v1
API key — your ak_... key
Models — Kilo auto-fetches the list; select what you want and Submit

2b. CLI (`kilo.json`)

Put the provider block in your global config (~/.config/kilo/kilo.json) — {env:...} references only resolve in trusted config and are silently ignored in a project-level ./kilo.json.

export CORALBRICKS_API_KEY="ak_..."

{
  "provider": {
    "coralbricks": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "CoralBricks",
      "options": {
        "baseURL": "https://inference.coralbricks.ai/v1",
        "apiKey": "{env:CORALBRICKS_API_KEY}",
      },
      "models": {
        "glm-5.2-fp4": { "name": "GLM 5.2", "limit": { "context": 1048576, "output": 32768 } },
        "kimi-k3": { "name": "Kimi K3", "limit": { "context": 1048576, "output": 32768 } },
        "kimi-k2.6": { "name": "Kimi K2.6", "limit": { "context": 262144, "output": 32768 } },
        "gpt-oss-120b": { "name": "GPT-OSS 120B", "limit": { "context": 131072, "output": 32768 } },
      },
    },
  },
  "model": "coralbricks/glm-5.2-fp4",
}

Available models

Model	Model ID	Context	Input $/M	Output $/M
GLM 5.2	`glm-5.2-fp4`	1M	$1.40	$4.40
Kimi K3	`kimi-k3`	1M	$3.00	$15.00
Kimi K2.6	`kimi-k2.6`	256K	$0.76	$4.00
GPT-OSS 120B	`gpt-oss-120b`	128K	$0.15	$0.60

Cached input tokens are always free — and Kilo's agent loop re-sends its context on every turn, which is exactly where cached-input pricing matters.

Notes

Anthropic Messages also works — pick Anthropic Messages as the Provider API with base URL https://inference.coralbricks.ai to run on our native Anthropic-format endpoint instead.
Tool calling, streaming, and reasoning work out of the box across the model line.
The same key works in OpenCode, Cursor, Cline, and Claude Code. Full API surface at /docs.