# Using Claude in Kilo Code

Kilo Code is an agentic coding extension in the Cline/Roo lineage with pluggable providers. Point it at AI Prime Tech and its plan-and-edit loops run on Claude — with the freedom to switch to another model the gateway serves when a task calls for it.

## Choosing a provider

In Kilo's settings, pick the API Provider. For the gateway's OpenAI-compatible surface, choose OpenAI Compatible, set the Base URL with /v1, paste your key, and type the model id. For native Anthropic, choose Anthropic and tick Use custom base URL with the host root.

Anthropic, Claude Code, and OpenAI-Compatible are separate provider flows in Kilo — match the one to your gateway's wire format.

## Model id and context settings

In OpenAI Compatible mode the Model ID is free-text and unvalidated, so a typo silently 404s as 'model not found' — enter exactly what the gateway exposes. Under Model Configuration, set Context Window and Max Output Tokens yourself or Kilo may mis-budget and truncate.

Because AI Prime Tech is multi-model, that free-text id is also how you point Kilo at a different model without changing providers.

## Tool calling for agentic edits

Kilo's edits, command runs, and MCP depend on the model and gateway supporting tool calls. Enable the Computer Use / function-calling capability in OpenAI Compatible mode if the gateway forwards tool calls, and Image Support only if it passes vision through.

These capabilities aren't auto-detected for a custom gateway, so set them to match the underlying model.

## Sustained throughput

Agentic plan-and-edit loops carry full-file context and a tool call per step, so token throughput stays high through a coding session — the same pattern as Cline and Roo.

A gateway gives you one billing surface for all that throughput plus the option to drop to a cheaper model for routine edits.

## Example

```
# Kilo Code -> Settings -> API Provider
# OpenAI Compatible:
#   Base URL: https://aiprimetech.io/v1
#   API Key:  <your key>
#   Model ID: claude-sonnet-4-5
#   Model Config -> set Context Window + Max Output
```

## FAQ

### How do I use Claude in Kilo Code?

Settings → API Provider → OpenAI Compatible (Base URL {SITE}/v1, free-text Model ID) or Anthropic (host root + Use custom base URL). Paste your gateway key.

### Why does Kilo say 'model not found'?

The Model ID in OpenAI Compatible mode is free-text and unvalidated — a typo 404s. Enter the exact id the gateway serves.

### Can Kilo reach non-Claude models?

Yes — through AI Prime Tech's multi-model gateway, the free-text model id can point at GPT or Gemini too.

### Do I need to set the context window?

Yes, in OpenAI Compatible mode set Context Window and Max Output Tokens manually or Kilo may truncate.

---

Source: https://aiprimetech.io/learn/use-claude-api-with-kilo-code/
Provider: AI Prime Tech — independent Anthropic-compatible API gateway (not affiliated with Anthropic).