Meta released Muse Spark Open, an 816B total/42B active parameter open-weight model with a 1M-token context window, designed for coding and agentic work.
Apple’s MLX framework now supports local and distributed inference for large language models on Mac Studio with Apple silicon.
Anthropic’s Claude Code team describes a shift from code review as a final gate to continuous AI-assisted review.
Stack Overflow for Agents saw signups triple after integrating a ChatGPT plugin.
Elorian launched Glimpse, its first AI model, as a research preview with API and waitlist-only chat access.
Beam is a 501B-parameter sparse Mixture-of-Experts model with 23B active parameters per token and a 1M-token context window.
Meta released Muse Spark Open, an open-weight frontier multimodal mixture-of-experts model with approximately 816B total parameters and ~42B active per token.
Apertus 1.5 extends the 8B and 70B open-weight models with native image/audio input, a reasoning 'thinking mode,' and a 262,144-token context window.
Gemini 4 Argon is available globally for developers, enterprises, scientists, and consumers via API in AI Studio and the Gemini platform, Google AI Ultra subscribers in Antigravity, and AI Ultra/Pro subscribers in the Gemini app and AI Mode in Search.
Falcon ASR is a new 1.6B-parameter speech recognition model for Arabic, with emphasis on the Emirati dialect.
OpenAI suspended accounts tied to two influence operations that used its models to fabricate personas and organizations for geopolitical messaging.
Nemotron models fine-tuned to surpass gold thresholds in IOI 2026 and IMO 2026 competitions.
Playground lets users create, play, and share custom games using only text prompts, with no coding required.
OpenAI disclosed new results on open problems in mathematics derived from an internal frontier model.
A new analysis argues that current cyber-risk discourse ignores trade-offs between open and closed models, and misrepresents China’s risk posture.
Mistral AI released Mistral Large 4 (ML4), a 1 trillion-parameter multimodal model.
Falcon-Emirati-7B is a new 7B-parameter model specialized for Emirati Arabic, built on the Falcon-H1-Arabic base.
Z.ai’s GLM 5.3, a 753B-parameter mixture-of-experts model, is now available on Amazon Bedrock.
OpenAI details its planned approach to EU text provenance requirements using watermarking technology.
A local evaluation of the Qwen3.8 27B model tested its ability to add integers and return answers in English words across 5,070 cases.
pwasm 0.2a0 is a pure-Python WebAssembly engine released by Simon Willison.
Anthropic’s Frontier Red Team evaluated GLM-5.3 and Claude Mythos Preview on 100 randomly selected binary exploitation tasks.
Chatham Financial is using OpenAI's Codex and a newer GPT model to automate capital markets workflows.
Google AI detailed a slate of AI updates in September 2026, including the frontier model Gemini 4 Argon with a 1-million-token output limit and advanced reasoning capabilities.
OpenAI says it disrupted a coordinated campaign aimed at extracting protected model reasoning through adversarial distillation techniques.
Google DeepMind introduced Gemini 4 Argon, a frontier AI model designed for complex workflows in software engineering, enterprise knowledge work, and cybersecurity defense.
Gemini 4 Argon introduces an industry-leading 1 million token context window for sustained reasoning across complex, long-horizon workflows.
Gemini 3.8 Live with Live Avatar integrates real-time video streaming with speech for interactive enterprise agents.
OpenAI is granting the Government of Ukraine access to its Daybreak cybersecurity program to aid in defending civilian infrastructure.
Claude Opus 5.5 is positioned as the first in a new Claude 5.5 family, performing at the level of Claude Fable 5.1 for most tasks while costing 40% less to run than Opus 5.