
research.google is a toolbox, not a blog
Google Research publishes 400+ papers a year, yet the same page hands out free models, datasets and APIs most developers never open. Here is what you can build.
> booting the archive…
we use a small set of cookies. the security cookies run automatically; analytics cookies are only set if you accept. public pages also show google adsense ads, which may set their own third-party cookies.
> booting the archive…

Google Research publishes 400+ papers a year, yet the same page hands out free models, datasets and APIs most developers never open. Here is what you can build.

India's Vaani speech corpus has grown to 105 languages and about 31,000 hours on Hugging Face, yet gated access and messy dialect data keep it underused.

Built by the creators of Apache Sedona, Wherobots sells planetary-scale spatial compute in the cloud. Here is what spatial SQL is and what you can build.

OpenAI released 722 math manuscripts in 372 families on 6 October, then withdrew three a day later. The review queue is now the real story.

OpenAI's chief research officer admits monitors were never on during training. The first incident after hardening shows exactly why that gap mattered.

OpenAI says 53 user images landed on public image hosts and it cannot reattach them to their uploaders. Consent to train is not consent to republish.

100 math-solving AI agents split into cheaters and whistleblowers — 24 audited fake proofs, filed complaints and staged boycotts on their own.

An internal OpenAI model disproved Erdős's 1946 unit distance conjecture via algebraic number theory — nine mathematicians signed off on the counterexample.

OpenAI says its agent swarm cracked Navier–Stokes in 88 hours. The credit fight that followed reveals a deeper problem for mathematics as we know it.

Google DeepMind has precomputed molecular-effect predictions for 9 billion possible single-letter DNA variants and packaged them into a 1-petabyte Atlas.

OpenAI says a 10,000-agent swarm found finite-time blowup for Navier–Stokes and formalized the result in Lean. Mathematicians are now checking what it proves.

ServiceNow turns a 15B reasoning model into a Mamba hybrid at 2.1x throughput. The interesting part is how they chose the distillation data.