TokenRouter is a unified, living catalog of the world's models — filterable by capability, modality, context, and cost — backed by deep research. Stop choosing models on vibes.
The model landscape shifts monthly. TokenRouter keeps the whole field in one searchable, filterable view — so your choice is evidence, not habit.
Live-ingested from OpenRouter, Google, Anthropic, Groq, xAI, and Hugging Face into a single searchable master catalog — commercial APIs and open-source side by side.
Filter by capability, input and output modality, context window, provider, reasoning models, free-only. "Vision plus tool-use under a dollar?" is one query, not an afternoon.
Cards auto-tiered Free / Budget / Standard / Premium from live per-million-token pricing — see what a model costs before you commit to it.
Models vetted and organized into collections for real use — a curation workflow, not just a list.
TokenRouter treats model selection as a real decision — made with data, not a default you set once and forget.
Behind the catalog: a structured capability framework and a research corpus spanning eleven generative-AI domains — text, image, video, audio, voice, music, 3D, and more.
Find the cheaper model that actually fits — pricing tiers at a glance, before the bill teaches you.
Compare providers on real capability data before committing — evidence, not marketing claims.
Find models by input and output modality — vision, audio, video, embeddings — across every provider at once.
The landscape shifts monthly. A live-ingested catalog means your shortlist never goes stale.
Model Engineering — our custom training and advisory service — designs your routing strategy, runs evaluations on your real tasks, and handles fine-tuning and cost optimization, on TokenRouter's evidence.
Browse the catalog today, or let us engineer your model strategy.