Thinking Machines

Tinker is now generally available. We also added support for advanced vision input models, Kimi K2 Thinking, and a simpler way to sample from models. thinkingmachines.ai/blog/tinker-ge…

Guest post by @_Mantic_AI on training LLMs to predict world events in Tinker thinkingmachines.ai/news/training-…

We’re thrilled to partner with @thinkymachines to deploy at least 1 gigawatt of NVIDIA Vera Rubin systems for frontier AI model training.

We are partnering with @nvidia to power our frontier model training and platforms delivering customizable AI. thinkingmachines.ai/news/nvidia-pa…

Four Qwen 3.5 models from @Alibaba_Qwen are now live on Tinker. Qwen 3.5 introduces hybrid linear attention that enables long context windows, as well as native vision input.

Custom fine-tuning has boundless applications. This week’s roundup shows Tinkerers taking models in unexpected directions, from science research to fun and games (literally).

Our second roundup of community projects highlights all things RL, from tutorials to APIs to cutting-edge research.

Kimi K2.5 is live on @tinkerapi:

Tinker now has a dedicated home on X 🤖🏡

@axiommathai To be specific, AxiomMath used Tinker to do RL in developing AxiomProver!

Congratulations to @axiommathai on their achievement! AxiomProver, a mathematics model fine-tuned with Tinker, got top scores on the Putnam Math Competition.

Science is best shared! Tell us about what you’ve built or discovered with Tinker, so we can tell the world about it on our blog. More details at thinkingmachines.ai/blog/call-for-…

In addition to expanding capacity, we are adding new models to our lineup and working on image support and production inference. We are excited to see what you build with Tinker!

Starting Monday, November 3rd, Tinker is switching to a pricing plan that reflects compute usage. This will ensure we have sufficient capacity to clear our waitlist by the end of the year, allowing anyone to sign up and start Tinkering. tinker-console.thinkingmachines.ai/rate-card

Roadmap update: Tinker launched into private beta a month ago, and we've seen hundreds of builders and researchers train and experiment with models on our platform. This month we've added new models, expanded the cookbook, and improved overall capacity and performance.

Today we’re announcing research and teaching grants for Tinker: credits for scholars and students to fine-tune and experiment with open-weight LLMs. Read more and apply at: thinkingmachines.ai/blog/tinker-re…

We just added 4 new models to Tinker from the gpt-oss and DeepSeek-V3.1 families. Sign up for the waitlist: thinkingmachines.ai/tinker/

Our latest post explores on-policy distillation, a training approach that unites the error-correcting relevance of RL with the reward density of SFT. When training it for math reasoning and as an internal chat assistant, we find that on-policy distillation can outperform other approaches for a fraction of the cost. thinkingmachines.ai/blog/on-policy…

Tinker is cool. If you're a researcher/developer, tinker dramatically simplifies LLM post-training. You retain 90% of algorithmic creative control (usually related to data, loss function, the algorithm) while tinker handles the hard parts that you usually want to touch much less often (infra, forward/backward of the LLM itself, distributed training), meaning you can do these at well below <<10% of typical complexity involved. Compared to the more common and existing paradigm of "upload your data, we'll post-train your LLM", this is imo a more clever place to "slice up" the complexity of post-training, both delegating the heavy lifting, but also keeping majority of the data/algorithmic creative control. I think the community still has to discover how and when finetuning makes sense compared to the (often strong) baseline of prompting a giant model. The early indications I've seen is that finetuning isn't so much about "stylizing" an LLM, instead, it's a lot more about narrowing the scope, and especially when you have a lot of training examples. An extreme example of scope narrowing being that of categorical classifiers, e.g.spam filters, content filters, etc. but it should be broader than that. Instead of building a giant few-shot prompts for a big LLM, it might work a lot better (and faster!) to finetune a smaller LLM specifically for your narrow task. Increasingly, production applications of LLMs are larger pipelines where a bunch of LLMs collaborate in DAGs and flows. Some of these components might work well as prompts. But a lot of it will probably work a lot better as a finetune. Tinker makes the latter trivial and should allow for an easy experimentation of what works best at any stage.

Very excited to see the Tinker release by @thinkymachines! @robertnishihara and I had a chance to experiment with the API, see anyscale.com/blog/fine-tuni…. It does a nice job of providing flexibility while abstracting away GPU handling. This will be 🔥 when combined with @raydistributed for simulations, inference and data processing. Looking forward to all the experimentation this unlocks! anyscale.com/blog/massively…