Chinese AI startup DeepSeek has released a significant update to its open-source reasoning model, bringing it closer to competing with paid services from OpenAI and Google. The new DeepSeek-R1-0528 model shows substantial improvements in complex reasoning tasks across mathematics, science, and programming. VentureBeat’s Carl Franzen reports that the updated model achieved 87.5% accuracy on the …
Google unveiled Gemini Diffusion at its I/O developer conference, marking a significant shift in how AI models generate text. The experimental model uses diffusion technology instead of the traditional transformer approach that powers ChatGPT and similar systems. The key advantage is speed. Gemini Diffusion generates text at 857 to 2,000 tokens per second, which is …
Prime Intellect has introduced INTELLECT-2, a 32B parameter AI model that represents the first of its kind trained using globally distributed reinforcement learning. The model employs a decentralized approach, utilizing compute resources from contributors around the world rather than centralized GPU clusters. In a technical report, Prime Intellect details their custom-built infrastructure components, including PRIME-RL, …
Anthropic has launched a new “Max” subscription tier for its Claude AI assistant, offering up to 20 times higher usage limits than its Pro plan. According to the announcement from Anthropic today, the new plan is designed for users who collaborate extensively with Claude and need expanded access for demanding projects. The Max plan comes …
Google has announced that its Gemini 2.5 Pro model is now available in public preview through the Gemini API in Google AI Studio, with Vertex AI rollout expected shortly. According to Google, this model is their “most intelligent” to date and has been priced competitively at $1.24 per million input tokens and $10 per million …
Midjourney, primarily known for AI image generation, has released new research in collaboration with New York University on training large language models to produce more creative text. Carl Franzen reports for VentureBeat that the research introduces two new techniques: Diversified Direct Preference Optimization (DDPO) and Diversified Odds Ratio Preference Optimization (DORPO). These methods encourage LLMs …
Poe, a platform for exploring and comparing AI models, has released its “Early 2025 AI Ecosystem Trends” report revealing significant shifts in user preferences across text, image, and video generation models. According to the report, OpenAI and Anthropic dominate text generation with approximately 85% of message share, while newcomers like DeepSeek and Google’s Gemini are …
Cohere’s research division has launched Aya Vision, an open-weight vision model supporting 23 languages. According to Carl Franzen’s report in VentureBeat, the model comes in 8-billion and 32-billion parameter versions and can analyze images, generate text, and translate visual content. Aya Vision outperforms larger models like Llama 90B while requiring fewer computational resources. The model …
Microsoft has launched a native Copilot app for macOS users in the US, UK, and Canada. According to Tom Warren from The Verge, the app provides access to Microsoft’s web-based AI assistant, allowing users to generate images and text or upload images. The Mac version includes dark mode support and can be activated with Command …
Inception Labs has unveiled Mercury, a new family of diffusion-based large language models (dLLMs) that can generate text up to 10 times faster than conventional autoregressive LLMs. According to the company, Mercury models can process over 1,000 tokens per second on NVIDIA H100 GPUs, speeds previously achievable only with specialized hardware. The company’s first publicly …