tinker-nomics

Interactive RLVR cost model for Tinker

tinker-nomics tinker-nomics Tinker-nomics The economics of post-training By Simon Guo · March 2026 🧮 Tinker vs Self-hosted RL Cost Calculator Pick model, workload, steps, batch size — get a side-by-side breakdown. → Lately, I have been Tinkering a lot and learning about post-training: I can easily build RL pipelines, train on big models that I would have never been able to fit on my tiny Stanford GPUs, plus I got some free credits through the research grant program. . I keep getting the same questions when I try to get my labmates to use it or convince advisors to pay for it: “ Why not just r

saved by

• Sebastian J Zhao

related reading

• On-Policy Distillation - Thinking Machines Labthinkingmachines.ai
• Tinkerthinkingmachines.ai
• Anatomy of a Modern Finetuning APIbenanderson.work
• Cheap RL tasks will waste compute | Mechanize, Inc.mechanize.work
• Spending Inference Time - Kevin Lukevinlu.ai
• Inkling: Our Open-Weights Model - Thinking Machines Labthinkingmachines.ai
• How can LLM RL Work Despite Information-Theoretic Inefficiencyberen.io
• [2510.13786] The Art of Scaling Reinforcement Learning Compute for LLMsarxiv.org
• Training LLMs to Predict World Events (Guest Post with Mantic) - Thinking Machines Labthinkingmachines.ai
• Keep the Tokens Flowing: Lessons from 16 Open-Source RL Librarieshuggingface.co
• Transformer Math 101 | EleutherAI Blogblog.eleuther.ai
• Why compute might get 10x more expensive in coming yearsdwarkesh.com