Simon Willison news
The latest AI and developer news from Simon Willison, with short summaries.
Visit website →GPT-5.4 Cyber
OpenAI has released GPT-5.4-Cyber along with an identity-verified cyber access program.
AI Security: Defense Strength Is Determined by Token Investment
AI-powered security review becomes more effective in proportion to the amount of tokens invested, changing the economic paradigm of security.
Google AI Debate
Steve Yegge's criticism of Google's AI adoption was directly rebutted by people inside Google.
Gemma 4 Transcription
An example of transcribing Gemma 4 E2B audio on macOS using MLX and mlx-vlm.
Monthly LLM Briefing
Simon Willison introduces a $10/month sponsor subscription and a monthly LLM briefing.
Voice Mode Was Weak
ChatGPT's voice mode doesn't use the latest model, but a weaker one from the GPT-4o era.
Check repository size
You can instantly check a repository's size via the GitHub API and share it by URL.
Muse Spark Unveiled
Meta unveiled Muse Spark and revealed the 16 tools of the meta.ai chat.
Z.ai Releases GLM-5.1, a 754B-Parameter Open Model
China's Z.ai has released GLM-5.1, an open-weight model with 754B parameters.
Anthropic Unveils Security Model Mythos
Anthropic has released Claude Mythos in a limited launch for security professionals, equipped with powerful cybersecurity capabilities.
Google Launches App for Running Gemma 4 on iPhone
Google has launched 'Google AI Edge Gallery,' an official app that lets users run Gemma 4 models locally on iPhone.
Surge in Kernel Security Reports Due to AI Slop
Linux kernel security reports are surging due to low-quality AI-generated reports.
Google unveils multimodal Gemma 4
Google DeepMind has released the new Gemma 4 series of open models supporting vision and audio.
Gemini 3.1/Gemma 4 Support
The llm-gemini plugin has been updated to support Gemini 3.1 and Gemma 4 models.
datasette-extract adds LLM management features
datasette-extract 0.3a0 introduces model configuration management via datasette-llm.