Benchmarks - Page 15
Stories from Benchmarks come first, newest first, then stories from connected topics (AI Models, LLM, AI, coding agents, preprint, open weights), each labelled with where it comes from. This view is not indexed by search engines.
Latest
Deep Principle Opens Hydrogen-1.0, a 753B-Parameter Science Model
Hydrogen-1.0 is an open-weight science model built on GLM-5.3. Deep Principle's own tests claim gains in chemistry tasks; no independent results yet.
Tencent Agent Trains Two Open Protein Models, HyInv and HyPLM
Tencent Hunyuan has released HyInv and HyPLM-Flash, two protein models it says its Hyra 2.0 research agent trained, with open weights under an MIT licence.
Grok API Text Will Carry an Invisible Watermark From 7 November
The Grok API watermark arrives on 7 November: text from seven Grok models gets an invisible mark by default, citing the EU AI Act. Admins can opt out.
Nesformer Compiles an NES Processor Into a Qwen-Style AI Model
Nesformer compiles the NES's 6502 processor into the weights of a Qwen3.5-style model with no training. What that means, and the caveats.
DeepSeek Open-Sources a Superagent That Works Inside a GitHub Repo
DeepSeek superagent prototype is out under MIT: a research coding agent that takes requests on GitHub or Feishu and replies with pull requests.
SenseTime Open-Sources SenseNova Loop, a 9B Model That Reuses Layers
SenseNova Loop, SenseTime's new open 9B model, runs a 24-layer block twice with shared weights. It is Apache 2.0, with no technical report yet.
Google Is Shutting Jules; Sessions End 11 November, Export Opens Now
Jules shutdown: Google's coding agent goes read-only after 11 November and closes on 31 December. How to export sessions, memories and settings now.
Pika Opens H3 Flex Video Weights, but Licence Excludes UK and US
Pika's H3 Flex video model, built on MiniMax H3, has downloadable weights, but its licence excludes the US, UK, EU and South Korea.
OpenAI Rebuilds Its Model Spec Around Five Core Values for GPT
OpenAI has reorganised its Model Spec around five values, adding rules on human oversight and context and dropping most of its style guidance.
Meta Says LLM-Written Tests Helped Catch 74 Serious Privacy Bugs
Meta says its LLM-generated JiTTests, built to fail on risky code changes, helped catch 74 serious privacy bugs before they landed in H1 2026.
Latest on TSN
Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.
