Meta Unveils Muse Wonder Models in Bid to Reclaim the AI Frontier
Meta introduced Muse Wonder and Muse Wonder Full, frontier-class models designed to compete with OpenAI and Anthropic while powering the company’s Muse agent.
LLMsConfirmed76
Meta introduced Muse Wonder and Muse Wonder Full, frontier-class models designed to compete with OpenAI and Anthropic while powering the company’s Muse agent.
OpenAI introduced GPT-6 Sol and Luna on September 22, pairing stronger reasoning with lower operating costs for coding, agents, and everyday work.
Google released Gemini 3.8 Flash, a 1-million-token model priced from $0.75 per million input tokens and built for coding, agents, and enterprise workflows.
Google launched Gemini 3.7 Flash three weeks after Gemini 3.6, while its delayed 3.5 Pro model keeps pressure on the company’s AI development strategy.
Google says Gemini has reached 1 billion monthly users, weeks after estimates placed ChatGPT at the same milestone, intensifying the consumer AI race.
OpenAI is removing text chat rate limits for ChatGPT Free and Go users starting next week and making the new GPT-5.6 Luna model the default, the company announced on August 6.
Meta released its first AI coding agent, Muse Code, alongside an upgraded Muse Spark 1.1 model with aggressive pricing, stepping into direct competition with OpenAI and Anthropic for enterprise developer dollars.
Reddit is launching Rules Hub, a suite of AI moderation tools that use large language models to evaluate whether posts match the intent of community rules, starting with new subreddits before expanding site-wide later this year.
Google will begin removing Assistant from Android phones, tablets, and paired wearables starting September 4, forcing a complete transition to Gemini as the sole AI assistant.
Apple's rebuilt Siri AI in iOS 27, launching this fall, instantly becomes the world's most widely distributed chatbot by tapping into personal iPhone data, but it enters a market where ChatGPT has already evolved into a more powerful agent.
Alibaba released Qwen3.8-Max, a 2.4 trillion parameter flagship AI model with benchmark scores it claims rival Anthropic’s Claude Fable 5 and surpass OpenAI’s GPT-5.6 Sol, intensifying the US-China AI arms race.
DeepSeek's re-post-trained V4 Flash 0731 release beats the much larger V4-Pro-Preview on agent benchmarks while costing as little as $0.0236 per complete agentic task, resetting the price-performance equation for open-weight models.
Google fixed 1,072 Chrome security bugs across two June updates, surpassing the previous 23 versions combined, as AI vulnerability hunting forces a new twice-a-week patching cadence.
Moonshot AI released Kimi K3, a 2.8-trillion-parameter open-weight model that rivals proprietary frontier systems, trained on roughly 20,000 Nvidia Hopper and Blackwell chips accessed through foreign cloud providers in defiance of US export controls.
Anthropic released Claude Opus 5 on Friday, a heavyweight model that outperforms the restricted Fable 5 on several benchmarks while being cheaper and free.