A 1,048,576-token window and 262,144-token answers, at $0 per token on OpenRouter.
Ling 3.2 Flash is inclusionAI's hybrid reasoning model — the flash tier from Ant Group's AI lab — and it reads text and images, up to 200 images per request. Thinking runs by default and can be switched off per call.
The catch: only the :free variant is free, it runs on a single provider (Ant Digital, fp8), OpenRouter's free-model request limits apply, and no end date is published.
How to claim: 1) Sign up at openrouter.ai → 2) Create an API key → 3) Send requests to inclusionai/ling-3.2-flash:free.
What changed
OpenRouter listed Ling 3.2 Flash as a free model on Nov 13, 2026. The listing gives a 1,048,576-token context window, a 262,144-token maximum reply, text and image input, and $0 per million input and output tokens. Requests are served by Ant Digital's fp8 deployment, with tool calling, structured outputs and response formats supported.
What it means for you
At $0 a token, the limits are the budget: one request can carry 1,048,576 input tokens — the same window as the Gemini 4 Hydrogen free tier we covered on Nov 13, 2026 — and return up to 262,144. That covers a long codebase, a run of screenshots or a day of agent logs in a single call. A free OpenRouter account is enough to call it; the free-model request caps, not a balance, set how much you get through. The catalog entry is /offer/openrouter-free-models.
Related free offers
- OpenRouter Free AI Models — free
:freechat, embedding and rerank models at $0. - BAI Limited Free LLM Models & Developer API Access — free Hy3, Mimo2.5, GLM 5.3 Flash and Qwen3.8 Flash.
- Agnes AI Free API — free OpenAI-compatible API for text, image and video Flash models.