Guide · updated 2026-09-26
Free models
The 13-model free pool, the auto-router, and how to pick.
Seventeen open models at $0 each, plus an auto-router and a keyless baseline — nineteen free ways to get an answer.
What is in the pool
Qwen 3.8 27B, both Gemma 4 sizes, Nemotron Lightning, Nano, Super and Ultra, both Poolside Laguna code specialists, GLM 5.2, Dots 3, both Ling Flash flavors, LFM 2.5, and both Inkling models, Cohere North Mini Code — plus a keyless baseline that answers even when every provider is busy. The workbench, chat, CLI, and desktop app all read the same live catalog from GET /api/models.php, so new models appear everywhere at once.
Auto-router first
Leave the picker on the default and requests ride openrouter/free: the best $0 model available right now. If it stumbles, the pool retries siblings automatically (up to 4 attempts), then falls back to the keyless baseline. Pick a specific model only when you want consistent behavior for a repeated task.
Every free model
openrouter/freeopenai-fastqwen/qwen3.8-27b:freegoogle/gemma-4-31b-it:freenvidia/nemotron-3.5-lightning:freenvidia/nemotron-3-nano-omni-30b-a3b-reasoning:freenvidia/nemotron-3-super-120b-a12b:freenvidia/nemotron-3-ultra-550b-a55b:freepoolside/laguna-s-2.1:freez-ai/glm-5.2:freedots-studio/dots-3-note-preview:freeinclusionai/ling-3.0-flash-fin:freeliquid/lfm-2.5-2.6b:freecohere/north-mini-code:freethinkingmachines/inkling:freeBrowse with pictures on the models page.
Proof on every reply
Each answer labels its exact model and path — sponsored for site-key $0 calls, free for the keyless baseline, your_key when your own provider key paid. If a label ever surprises you, the troubleshooting guide explains why.