Models & Labs
OpenAI launches GPT-5.6 and ChatGPT Work
The Verge AIhigh confidence
Why it matters
- →GPT-5.6's public release marks a significant advancement in AI model capabilities.
- →ChatGPT Work aims to make AI more accessible for non-technical users, enhancing productivity.
- →OpenAI's strategy could shift the competitive landscape in AI agents, challenging rivals like Anthropic.
OpenAI has launched GPT-5.6 to the public following approval from the Trump administration, ending a limited preview period. The company also unveiled ChatGPT Work, an AI agent that combines ChatGPT and Codex capabilities for non-coding tasks, powered by the GPT-5.6 model suite. This release aims to enhance productivity by allowing users to create documents, spreadsheets, and more, with integration into tools like Slack and Google Drive. The rollout is global, with desktop users gaining immediate access and a phased release for mobile and web users. OpenAI's move positions it as a strong competitor in the AI agent market.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
More from The Verge AI
More in Models & Labs
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
llama.cpp Releases
Models & Labscoding
llama.cpp b12342 adds few-row MMQ for speculative decoding
Speculative decoding just got a significant speed boost on NVIDIA GPUs. This release introduces a specialized kernel for 'few-row' matrix multiplications, targeting the small batch sizes typical of verification passes in speculative decoding workflows. By optimizing these specific operations for quantized formats like Q8_0 and IQ4 variants, llama.cpp reduces latency where it matters most during token generation. The change is CUDA-only for now but leverages hardware capabilities from Ampere onwards, making local inference faster without requiring new model architectures or complex configuration changes.
