Reflection AI Beam: New 501B Open-Weight Model
Reflection AI announced Beam, a 501B open-weight model trained on 10,500 GPUs to challenge Chinese labs. Public weights arrive by late October 2026.
Perplexity Moves Agent Sandboxes to NVIDIA Vera
Perplexity begins shifting agent sandboxes to NVIDIA Vera CPUs, aiming to cut single-threaded execution bottlenecks in automated reasoning loops.
Perplexity open-weight models target routing and search
Perplexity released open-weight models pplx-decider and pplx-embed to cut costs for agent workflows and contextual vector search systems.
OpenAI Agents API adds hosted browser and sandboxes
OpenAI Agents API update introduces hosted browser automation, persistent sandboxes, and GPT-6.1 Sol support. Service remains in public beta.
OpenAI Adds textGrain Watermarks to ChatGPT in the EU
To comply with the EU AI Act, OpenAI activates textGrain watermarks for ChatGPT only in Europe, while keeping the detection tool restricted to researchers.
Fine-Tuning Nemotron 3.5 ASR for Regional Accents
NVIDIA published an open guide to adapt Nemotron 3.5 ASR to regional dialects in 4.5 hours while avoiding catastrophic forgetting.
NVIDIA Speeds Up GPT-6 Astra Ultrafast on Blackwell
NVIDIA and OpenAI boost GPT-6 Astra Ultrafast on Blackwell GPUs. Output speeds jump 8x, but overall latency drops 2.6x at $300 per million tokens.
Cursor adds GLM 5.3 and GLM 5.3 Flash models
Cursor adds native support for Z.ai's GLM 5.3 and GLM 5.3 Flash models, featuring a 1M token context window and on-demand pricing outside standard plans.
Meta Muse Charm: Pocket AI Device Bypasses Phones
Meta unveiled the Muse Charm, a pocket AI keychain with a camera and quick fingerprint wake-up, aiming for hardware beyond Apple and Google app stores.
Training AI agents across tools stops interface lock-in
Hugging Face shows how multi-harness reinforcement learning stops AI coding agents from breaking when moved between developer tools.