Product blog

Works with popular AI services

Generative features use an OpenAI-compatible API. The examples below were exercised in development; availability and pricing depend on each provider. Configure yours in Settings → Model selection.

Llama 3.3 via OpenRouter

A powerful open-source model. Good choice if you want flexibility and no vendor lock-in.

Qwen via OpenRouter (free tier)

A capable free option to get started without an API bill.

Technical details (model IDs)

Moonshot / Kimi (Kimi-2.5): kimi-k2.5 (and equivalent K2.5 identifiers) via Moonshot’s OpenAI-compatible API.
OpenRouter — Llama 3.3 70B Instruct: meta-llama/llama-3.3-70b-instruct
Perplexity (Sonar): https://api.perplexity.ai — model ids such as sonar or sonar-pro
OpenRouter — Qwen 3.6 Plus Preview (free): qwen/qwen3.6-plus-preview:free — confirm current routing on openrouter.ai/models
OpenRouter — Xiaomi MiMO V2 Pro: xiaomi/mimo-v2-pro — confirm on openrouter.ai/models

Generative calls use a /v1/chat/completions-style endpoint for analysis, code assist, and related flows.

Other OpenAI-compatible providers may work; availability, pricing, and API quirks are determined by the vendor. Always test with your own API key and quotas.