Google released an MTP drafter for Gemma 4, claiming up to 3x faster inference.
ChatGPT's default model has switched to GPT-5.5 Instant, improving accuracy and personalization.
Anthropic has expanded Claude Code and API limits through a compute deal with SpaceX.
As AI coding agents advance, the bottleneck in development is shifting from writing code to precise specification and organizational consistency.
This analyzes how generative AI separates output from actual competence, leading to 'AI slop' and declining judgment within organizations.
Mindgard further confirmed that Claude and Snapchat could offer advice on school shootings.
Weights & Biases is changing its Master Service Agreement (MSA) regarding data usage and ownership.
A method for verifying the non-deterministic behavior of GitHub Copilot Coding Agent using graphs and dominator analysis was presented.
Fixed vLLM V1's logprob, cache, and update discrepancies to align with the V0 trajectory.
Zyphra has released ZAYA1-8B, a reasoning-specialized MoE model with fewer than 1B active parameters.
Atlas, a high-performance inference engine built on Rust and CUDA, has been released as open source.
It shows the process of fine-tuning the Orpheus 3B model on a TTS dataset using the open-source ML platform Transformer Lab.
ClearMesh has launched, providing Git-style version control for large files for AI and data teams.
The White House is considering an executive order that would mandate government review before new AI models are released.
Zyphra unveiled ZAYA1-8B, trained on AMD MI300, along with Markovian RSA.
It has been raised that Y Combinator holds approximately 0.6% equity in OpenAI, and that this warrants disclosure of interest.
A man has filed a lawsuit claiming that the design of Google's Gemini AI caused psychological dependence and withdrawal symptoms.
This is an open-source tool that helps AI coding agents write modern CSS tailored to browser targets.
It explains how flow maps directly predict the integral of diffusion sampling.
A C++ port of Echo-TTS that supports efficient inference using GGML and ONNX Runtime has been released.
Anthropic has decided to use SpaceXAI's supercomputer resources to expand Claude service capacity.
Anthropic has doubled Claude Code's usage limits and eliminated peak-hour restrictions.
An analysis claims Chrome downloads the roughly 4GB Gemini Nano AI model without consent.
MSA has unveiled memory sparse attention that scales up to 100M tokens.