Tag
Phoenix now supports Meta's Muse Spark 1.3, a model released quietly alongside others, and it deserves more attention than it received.
Meta introduces Muse Voice Transcribe, a real-time audio perception model that excels in streaming ASR and diarization with multilingual support, topping public benchmarks.
John Kim, an L7 Senior Staff SWE at Meta, recaps an AMA session discussing topics like the future of software engineering jobs, AI coding workflows, and personal experiences in the tech industry.
A step-by-step guide to fine-tuning MetaAI's Muse Glimmer model locally without cloud dependencies.
Meta AI has launched a new Mac app with system-wide dictation and context-aware features powered by its Muse Spark model, offering business integrations for insights and task automation.
Meta has launched an AI desktop app for macOS with features like screen sharing and system-wide dictation, aiming to compete with other AI assistants such as ChatGPT and Claude.
Spencer Barnett announces the launch of the Meta AI Mac OS app, highlighting its dictation feature that allows high-accuracy voice input anywhere on the computer.
Meta has launched a dedicated Mac app for its AI chatbot, enabling screen sharing for suggestions, cross-app dictation, and enhanced business tools integrated with social media and Google Workspace.
Meta's Muse Glimmer is a 30B multimodal Transformer model designed for autonomous agentic tasks on consumer hardware, using a memory hierarchy and quantization to fit within 24-32 GB envelopes.
The article describes extending the Muse Glimmer 30B model's context to 512k tokens using its unique architecture with RoPE in SWA layers and no position encoding in GQA attention layers, with benchmark results showing high performance up to 512k tokens.
A new Meta paper reveals that small models can accurately predict scaling laws but require more extensive hyperparameter tuning. The study finds scaling laws emerge around 4M parameters and become clearer with proper tuning.
Zuckerberg is pitching AI as the future of Meta's ad revenue, but their AI ad tools are generating errors like turning dresses into pants, amid broader industry backlash and new regulations in AI advertising.
Meta's Muse Spark 1.1 update upgrades Meta AI to connect with email and calendars, create slides, and run recurring tasks—shifting from one-off conversations to persistent, cross-app activity with new concerns around permissions and control.
MARCO is a Meta AI framework that decomposes clicks by intent to improve ads conversion prediction, correcting per-intent calibration bias and lifting conversions per click by +2.80% and topline metrics by +0.98% in production.
Meta releases Muse Code, a terminal coding agent, and Muse Spark 1.2, an upgraded coding-focused model with improved code generation, debugging, and long-horizon task handling.
A critical opinion piece arguing that Meta's AI strategy is unfocused and reactive, and that Zuckerberg's recent upbeat messaging on AI is a PR move to mask the company's lagging position.
Meta AI's new task-planning features raise concerns about conflicts of interest given Meta's advertising and commerce operations, prompting discussion of disclosure policies.
Mark Zuckerberg announced Meta's plans to push into personal AI agents during the Q2 2026 earnings call, aiming to create consumer-friendly agents that work out of the box, while also highlighting business agents on WhatsApp and Instagram.
The article examines how friction is being eliminated in digital experiences and AI companions while concentrated in physical infrastructure, creating three distinct economies of effort that reshape labor, loneliness, and value.
Meta is rolling out its Meta AI chatbot within Threads' DMs, allowing users to chat privately with the AI assistant, share posts, images, links, and videos, aiming to keep users within its ecosystem and discourage use of third-party assistants. Rollout begins globally from Monday.