Back to AI News Center
AI News Category
Product Launches
New AI tools, major product updates, and launch announcements.
112 published articles
Product Launches
Reflection AI Unveils Beam: A 501B Open-Weight Model Challenging Chinese AI on Inference Cost
On October 5, 2026, Reflection AI launched Beam, its first open-weight frontier model, a 501-billion-parameter sparse mixture-of-experts model designed to outperform Chinese open models on inference cost while matching Z.ai's GLM-5.2 on advanced reasoning benchmarks. Introducing Beam: A New Open-Weight Frontier Model Reflection AI's Beam model represents a strategic entry into the open-weight AI landscape. The company claims Beam matches the performance of Z.ai's GLM-5.2 on advanced reasoning tasks while...
Read Article
Product Launches
H Company Releases Holo4: Open-Weight Agentic Models for Generalist Computer Use
H Company has released Holo4, a new series of open-weight agentic models in 27B dense and 35B-A3B Mixture-of-Experts variants, along with an updated Holotron4 Nano.
Read Article
Product Launches
Fireworks AI's Ember-1 Slashes Reasoning Tokens by 40% While Outperforming Kimi K3
Fireworks AI's new Ember-1 model, built on Kimi K3, slashes reasoning tokens by ~40% at equal quality, outperforming Kimi K3-max on benchmarks and reducing cost per task by roughly half. It also set a new cost-vs-quality Pareto frontier on Doximity's Bedside Bench.
Read Article
Product Launches
Qualcomm Unveils Snapdragon 8 Elite Gen 6 & Extreme: On-Device AI, 8K Video, and Motorola Signature 27
Qualcomm's new Snapdragon 8 Elite Gen 6 and Extreme Gen 6 mobile platforms bring on-device agentic AI, 8K 60fps video, and a 30-billion-parameter MoE model to smartphones. The Motorola Signature 27 will be powered by the Extreme Gen 6, with general availability expected later this year.
Read Article
Product Launches
YouTube Unveils New AI Creator Tools at Made On YouTube 2026: Script Coaching, Gemini Editing & More
YouTube announced new AI creator tools at Made On YouTube 2026, including a storytelling assistant, dynamic thumbnails, and Gemini-powered editing for Shorts, alongside AI live translation and deepfake protection enhancements.
Read Article
Product Launches
Hugging Face Transformers Now Natively Supports GGUF Models for Local AI Inference
Hugging Face's Transformers library now natively supports llama.cpp GGUF quantized models, streamlining local AI inference and allowing developers to load efficient checkpoints from the Hugging Face Hub.
Read Article
Product Launches
OpenAI Upgrades ChatGPT Voice with App Integration and GPT-6 Models
OpenAI upgraded ChatGPT Voice on September 23, 2026, enabling it to access connected apps like Gmail and Slack for hands-free task completion. The update, powered by new GPT-6 models (Astra, Sol, Luna), is available globally in the ChatGPT app.
Read Article
Product Launches
Anthropic's Claude Opus 5.5: Fable 5.1 Performance at 40% Lower Cost with Enhanced Cybersecurity
Anthropic's Claude Opus 5.5, launched September 22, 2026, matches Fable 5.1 performance at 40% lower cost than Opus 5, featuring new cybersecurity safeguards and 20% reduced token prices.
Read Article
Product Launches
ByteDance Launches Dramagic AI Platform for Short Drama Production
ByteDance launched Dramagic on September 21, 2026, an AI platform for producing short dramas from script to screen. This enterprise solution, available through BytePlus, aims to reduce production costs and streamline content creation.
Read Article
Product Launches
Alibaba's Qwen AI Team Releases Qwen-Image-2.1 for Advanced Image Generation and Editing
Alibaba's Qwen AI team released Qwen-Image-2.1, a 7B open-weight image model for generation and editing, claiming it beats closed rivals. It supports transparent RGBA images and up to ten reference images, running on consumer GPUs like the RTX 3090.
Read Article
Product Launches
Grok 4.7 Benchmarks: How xAI's New Model Stacks Up Against Claude Fable 5.1 and GPT-6
xAI launched Grok 4.7 on September 21, 2026, positioning it as its most capable model for coding and knowledge work. While competitively priced, Grok 4.7 trails Claude Fable 5.1 and GPT-6 in overall intelligence but leads in specific coding and engineering tasks.
Read Article
Product Launches
Pirate Face Launches BitTorrent Service for Open AI Models, Enhancing Distribution Resilience
Pirate Face launched a new peer-to-peer service on September 20, 2026, mirroring open AI models from Hugging Face as checksum-verified BitTorrent torrents. This ensures model integrity and availability, even if original hosts remove them.
Read Article
Product Launches
Fujitsu to Launch 2nm FUJITSU-MONAKA CPU and Server for Sovereign AI Infrastructure in November 2026
Fujitsu will begin global sales of its 2nm FUJITSU-MONAKA CPU and Fujitsu MONAKA Server in November 2026. This made-in-Japan hardware offers world-class AI inference performance and power efficiency for sovereign AI infrastructure.
Read Article
Product Launches
Bend Language Blocks AI Coding Mistakes with Machine-Checked Proofs, Hits Hacker News
Bend, a new programming language, launched on September 18, 2026, to make AI-generated code trustworthy. It uses machine-checked proofs to verify code against declared rules before every commit, preventing common AI coding errors.
Read Article
Product Launches
PrismML Releases Bonsai 2 27B: A Highly Compressed LLM with Near-Lossless Performance
PrismML released Ternary Bonsai 2 27B, a highly compressed LLM that reduces Alibaba's Qwen3.8 27B model to 5.9 GB, achieving a 9-10x memory reduction with 98.2% performance retention. This update makes powerful AI models more accessible and efficient.
Read Article
Product Launches
Instinct Concierge and Meta's Muse AI Agents Now Offer Phone Call Capabilities
Instinct and Meta's Muse AI agents can now make phone calls for users. Instinct founder Noah Shinn announced "Instinct Concierge" for tasks like booking restaurants, while Meta's Muse added outbound calls to U.S. businesses.
Read Article
Product Launches
Snap's $2,200 AR Glasses Enhanced with New "Specs Intelligence" AI: What You Need to Know
Snap CEO Evan Spiegel introduced "Specs Intelligence," an anticipatory AI system, on September 16 to boost the appeal of its $2,200 AR glasses. This AI, launching in preview on iOS, learns user goals and provides relevant information, with Snap pledging strict privacy for personal data.
Read Article
Product Launches
Google DeepMind Launches Gemini 3.8 Live Models, Undercutting OpenAI on Price for Voice Agents
Google DeepMind launched Gemini 3.8 Live and 3.8 Live Extended Thinking on September 15, 2026. These new speech-to-speech models for voice agents offer advanced conversational AI at a lower cost than OpenAI's GPT-Live-1, making them a competitive option for developers and enterprises.
Read Article
Product Launches
Daydream Integrates Apple Intelligence for iOS 27 Fashion Shopping Features
Daydream launched new Apple Intelligence-powered features on September 14, 2026, enabling iPhone users to shop outfits from photos and search for clothing via Siri. These iOS 27 updates enhance fashion discovery with AI.
Read Article
Product Launches
Perplexity Brings Local "Portable Computer" AI Agent to Windows on NVIDIA RTX
Perplexity's "Portable Computer" AI agent is now available for Windows on NVIDIA RTX, allowing users to run local AI agent workflows with a post-trained Qwen 3.8 27B model. This update supports on-device processing for sensitive files and does not consume Perplexity Computer credits.
Read Article