Bugbot just received a massive upgrade making it over 3x faster, 22% cheaper, and 10% more effective at catching bugs, with 90% of runs now completing in under three minutes thanks to harness upgrades and the new Composer 2.5 model integration.
Anthropic launches two powerful Mythos-class AI models — Fable 5 for the public and Mythos 5 with lifted safeguards for cyberdefenders — pushing forward despite industry calls for a slowdown, backed by a $65 billion funding round valuing the company at nearly $1 trillion.
Apple is expanding its Private Cloud Compute infrastructure beyond its own data centers for the first time, partnering with Google Cloud and NVIDIA to run Apple Intelligence workloads while maintaining strict privacy and security guarantees through cryptographic verification and dual-vendor attestation.
Cohere launches North Mini Code, an open-source agentic coding model with a 30B mixture-of-experts architecture that uses only 3B active parameters, delivering up to 2.8x higher output throughput and 30% better latency than rival Devstral Small 2, now available on Hugging Face and via API under an Apache 2.0 license.
Google launches Gemini 3.5 Live Translate, a groundbreaking AI model delivering near real-time speech-to-speech translation across 70+ languages while preserving speakers' natural voice characteristics, now available to developers, enterprise users, and the public via Google Translate on Android and iOS.
Google's Magenta team releases Magenta RealTime 2, a cutting-edge open-weights AI music generation model available in 230M and 2.4B parameter sizes, enabling real-time music streaming on Apple Silicon Macs and offline inference on NVIDIA GPUs.
Roblox launches Cube, an open-source AI model family that generates 3D game assets, objects, and scenes directly from text prompts, with model weights publicly available on Hugging Face and trained on 2.8 million synthetic 3D assets.
Researchers from UIUC, UC Berkeley, and Chroma unveil Harness-1, a 20-billion parameter open-source AI search agent that outperforms GPT-5.4 on recall benchmarks while training on a fraction of rival data, achieving 73% recall by offloading memory management to a structured external environment.
Apple unveils free AI cloud access for smaller developers at WWDC 2026, offering no-cost Foundation Models via Private Cloud Compute to those with under 2 million App Store downloads, as skyrocketing AI costs plague the industry — with Uber reportedly burning through its entire 2026 AI budget in just four …
Xiaomi and TileRT claim a major AI speed breakthrough, hitting over 1,000 tokens per second on a trillion-parameter model using just 8 commodity GPUs — roughly 15 times faster than ChatGPT and Claude — powered by FP4 quantization and speculative decoding, with an open-source model checkpoint already live on Hugging …