Energy-Based Reasoners: 345x less inference compute for recurrent reasoning
A new reasoning architecture that learns an energy function over latent reasoning states and treats looped updates as gradient descent on that energy — 99.9% on Sudoku-Extreme with a 1.8M-parameter model.
Why it matters
The loop finally knows when to stop: an explicit energy measure of progress replaces blind iteration, giving state-of-the-art coverage of distinct valid solutions.
What you could build with it
A researcher could build a small on-device reasoner that iterates until its energy score says the answer is good enough, instead of paying for a giant model on every query.
Does it hold up?
Too early to judge: a research paper with strong benchmark numbers on algorithmic tasks, not a shipped model — no public weights or production use yet.
Built with Energy-Based Reasoners
- Energy-Based Reasoners: 385x more efficient inferencelaxima · Curated signal feed flagging the HN front-page run and the compute-efficiency claim.
- Wednesday, 18 November 2026 — Energy-Based Reasonerstokenreporter · Daily digest entry summarizing the architecture and the Sudoku result.
Learn more
First spotted on article: source.
More AI models
Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Qwen4-Max: Alibaba's new flagship model goes liveAlibaba's 2.4T-parameter flagship is now generally available via API, with a 1M-token context, native vision, and…models · JEV 0.67smolmaxxing: run 16B LLMs on a 1 GB TV box or a 2013 iPadAn MIT-licensed llama.cpp patch set that runs language models up to a 16B MoE on devices with about 1 GB of usable RAM…models · JEV 0.65Blue TTS v3: a 388MB open text-to-speech model with voice cloningAn open-source 388MB text-to-speech model (MIT) with voice cloning from about 10 seconds of speech, 7 languages, 15…models · JEV 0.65Haru-4BOpen full-duplex spoken-dialogue foundation model that listens and speaks simultaneously (7.5 Hz frames, 133 ms per…models · JEV 0.63
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.