Inkling and Inkling Small are now served directly by @thinkymachines on OpenRouter, free to use inside agentic harnesses only.
Plug them into Claude Code, Codex, Hermes Agent and more to unlock the free access.
Open-weight MoE reasoning models with native text, image, and audio input and a 1M context window. More on both below 🧵
https://t.co/wF46xTiiEW
@thinkymachines Inkling: 975B total parameters, 41B active. Configurable reasoning effort from minimal to max, with native image and audio understanding alongside text. https://t.co/i6jDkhV8nH
@thinkymachines Inkling Small: 276B total, 12B active. The efficient half of the family, with the same reasoning controls and the same multimodal input.
https://t.co/OsbdUrqCsU
@thinkymachines Note: Thinking Machines Lab logs prompts and outputs and uses them to improve their models, with session data disassociated from your account first. Full terms are on each model page.
The open weights are Apache 2.0. API access to these free endpoints runs on separate terms.
open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Why it matters
Free frontier-scale reasoning inside Claude Code and Codex lowers the cost of long-horizon AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → runs, with the tradeoff that Thinking Machines logs prompts and outputs to improve its models.
