AI Providers¶

Kaisho supports multiple AI providers. Configure them in Settings > AI or directly in settings.yaml.

Provider Overview¶

Provider	Prefix	Local/Cloud	Tool Calling
Ollama	`ollama:`	Local	Yes
Ollama Cloud	`ollama_cloud:`	Cloud	Yes
LM Studio	`lm_studio:`	Local	Yes
Claude (Anthropic)	`claude:`	Cloud	Yes
OpenAI	`openai:`	Cloud	Yes
OpenRouter	`openrouter:`	Cloud	Yes
Kaisho Cloud AI	`kaisho:`	Cloud	Yes

Ollama (Local)¶

Run models on your own hardware. Install Ollama, then:

# settings.yaml
ollama_url: http://localhost:11434
advisor_model: ollama:qwen3:14b

The UI auto-discovers available models from your Ollama instance.

Ollama Cloud¶

Use Ollama's hosted inference. Get an API key from ollama.com.

ollama_cloud_url: https://ollama.com
ollama_api_key: your-api-key-here
advisor_model: ollama_cloud:gemma4:31b

Note

Some models require a paid Ollama subscription. Free-tier models include gemma3:27b, gemma4:31b, qwen3-next:80b, deepseek-v3.2, and others. Check the Ollama Cloud dashboard for current availability.

LM Studio¶

Run models through LM Studio:

lm_studio_url: http://localhost:1234
advisor_model: lm_studio:your-model-name

Claude (Anthropic)¶

Use Anthropic's API directly. Get an API key from console.anthropic.com:

claude_api_key: sk-ant-...
advisor_model: claude:claude-sonnet-4-20250514

Warning

This requires a pay-per-use API key, not a Claude Pro consumer subscription. API and consumer accounts are separate.

OpenAI¶

openai_url: https://api.openai.com
openai_api_key: sk-...
advisor_model: openai:gpt-4o

OpenRouter¶

Access hundreds of models through a single API key from openrouter.ai:

openrouter_url: https://openrouter.ai/api/v1
openrouter_api_key: sk-or-...
advisor_model: openrouter:anthropic/claude-sonnet-4

Kaisho Cloud AI¶

When connected to Kaisho Cloud with a sync_ai plan, the advisor can use the hosted AI gateway. No API keys needed:

# Automatically configured when Cloud Sync is connected
advisor_model: kaisho:default

Separate Models for Advisor and Cron¶

You can use different models for interactive advisor queries and background cron jobs:

advisor_model: ollama_cloud:gemma4:31b
cron_model: ollama:gemma4:latest

This lets you use a cloud model for the advisor (better quality) and a local model for cron jobs (no API costs).

Checking Provider Status¶

In Settings > AI, the probe indicator shows which providers are reachable. Green means the provider responded; red means it's unreachable or the key is invalid.