Topic
Technique in which a language model writes intermediate steps as tokens before its final answer, placing its reasoning in the context window.
A dev.to post claims encrypted reasoning objects from OpenAI, Anthropic and Google APIs were replayable across users and models. The disclosure is thin, but the storage habit it exposes is yours.
Reality
NV-Reason-CT reads a 300-to-600-slice study as a single 3D volume instead of a stack of frames. NVIDIA published it as an uncleared research foundation whose cited multireader study measured its chest X-ray predecessor.
Reality
BDH-CQ updates a fixed-size internal memory instead of generating chain-of-thought tokens, and its authors put a single puzzle query at about $0.00070. Two outside researchers say the result does not yet credit the design.
Reality
A dev.to writeup credits researchers with decrypting 10,000 reasoning traces from OpenAI, Anthropic and Google models for about $720, and the storage question behind that claim sits in the API example the post itself prints.
Reality
Anthropic tested whether a model's visible reasoning names what actually drove its answer, and for two reasoning models it usually did not. Teams treating traces as an audit record are relying on that property.
Reality
Geiping's Huginn improves on reasoning benchmarks by looping a latent block instead of emitting chain-of-thought tokens, holding weights at 3.5 billion while per-token compute rises about fourteenfold.
Reality