Skip to content

Supported Providers

If you don’t have the workstation to host your own models, TomoriBot supports a wide range of LLM providers, image generation APIs, voice services, and search tools, as well as features to mix-and-match them. There are plans to add in more providers.

ProviderStreamingTool CallingImage InputEmbeddingsNotes
Google GeminiFree Models Available
OpenRouterFree Models Available
Anthropic (API)-NOT Claude Code
NovelAI--Only GLM 4.6 can use Tools
NvidiaFree Models Available
Deepseek---
Z.ai-Free Models Available
Z.ai Coding--Subscription Plan ⚠️ ToS restricts to coding/agent use only
Google Vertex AIIncludes ‘free’ Express version
Codex CLI (via ChatMock)-Instructions in setup-chatmock.md
ProviderText-to-ImageImage-to-ImageInpaintingNotes
Google--
OpenRouter--
NovelAICan be combined with other providers
Nvidia--
Z.ai---
ProviderText-to-VideoImage-to-VideoNotes
GoogleAsync polling workflow
OpenRouterAsync polling workflow
Z.aiAsync polling workflow
ProviderText-to-SpeechSpeech-to-Text
ElevenLabs

The LLM sees a single unified web_search(query, category) tool. A dispatcher routes each call through an engine chain (Brave → SearXNG → DuckDuckGo → Felo) and returns the first successful result. Individual engines are no longer LLM-visible.

For URL reading, the LLM sees fetch_url(url, max_length?, start_index?, raw?). It can route through an optional Crawl4AI sidecar first, then always falls back to internal mcp_fetch. It is unavailable on NovelAI.

EngineCategoriesIntegrationNotes
Brave Searchtext / image / video / newsREST APIFirst in chain when a Brave API key is configured. ⚠️ Set a $5 usage limit in the Brave dashboard to avoid surprise charges.
SearXNGtext / image / video / news / science / it / files / musicREST API (self-hosted sidecar)Self-hosted aggregator that proxies Google, Bing, DDG, Brave, Wikipedia, etc. See servers/searxng/README.md or simply run bun launch --searxng.
DuckDuckGotext onlyMCP serverDefault Web Search. Fallback when Brave/SearXNG are unavailable; transparently cascades to Felo on rate limits.
Felo AI Searchtext onlyMCP serverFinal-resort text fallback.
Crawl4AIURL fetchREST API (self-hosted sidecar)Browser-renders pages and returns markdown via /md; falls back to mcp_fetch when unavailable. See servers/crawl4ai/README.md or simply run bun launch --searxng
MCP FetchURL fetchBundled MCP serverDefault URL fetch. Mandatory final fetch_url fallback and default behavior when no browser sidecar is configured or healthy.