. . AI: Reasoning: Coming to a Model Near You Soon (4.10.24)

Friends, in today's AI news, we get some snippets of information on the next generation of foundational models. OpenAI and Meta both hint at models with further advanced reasoning capabilities.

OpenAI has launched a new version of GPT-4, which happens to also have a very respectable increase in mathematical reasoning, along with GPT-4 Turbo with Vision access opening up via API. We also distilled the major headlines covering open-source models, AI music generation, and AI longevity assistants.

Climate change: Interested in seeing how some in industry are talking about using AI to deal with climate change? Check out today’s GreenBiz conference panel AI for Nature, Climate and Beyond How New Tech Is Transforming Sustainability.

And now here’s today’s top AI news!

-Marshall Kirkpatrick, Editor

OpenAI and Meta Prepare New AI Models with Enhanced Reasoning and Planning Capabilities

First impacted: WhatsApp Users, Ray-Ban Smart Glasses Users
Time to impact: Short

The FT reports that OpenAI and Meta are preparing to launch new AI models, which they say will be capable of reasoning and planning. Meta plans to roll out Llama 3 in the coming weeks, while OpenAI, backed by Microsoft, indicates its next model, rumored to be called GPT-5, is coming soon. Reasoning is the crucial piece that today's models need to be significantly more capable, and the developments on this front are arguably the most important to track in the roadmap to AGI. [OpenAI and Meta ready new AI models capable of ‘reasoning’] Explore more of our coverage of: OpenAI, Meta, AI Reasoning. Share this story by email

Mistral AI Launches Mixtral 8x22B Model

First impacted: AI developers, AI researchers
Time to impact: Short

Mistral AI has launched Mixtral 8x22B, which it describes as its second-largest open-source model, according to a series of tweets. The model, an 8x22B MoE (Mixture of Experts), is an upgrade from the company's previous 8x7B MoE model, Mixtral. [opentrackr.org] Explore more of our coverage of: Mistral AI, Open-Source Models, Mixture of Experts. Share this story by email

OpenAI Launches GPT-4 Turbo with Vision API

First impacted: Developers, Data Analysts
Time to impact: Short

OpenAI has released a 'majorly improved' version of GPT-4 Turbo with improved reasoning capabilities, and GPT-4 Vision is now available through its API. The benchmark results for GPT-4 show a really impressive increase in results, especially in mathematical reasoning. The OpenAI team also shared some of the feedback and use cases from GPT-4V, which ranged from powering Devin, the AI software engineer, to estimating the caloric values of foods by sharing a visual with the model. Super impressive stuff! [via @OpenAI] Explore more of our coverage of: OpenAI, GPT-4 Turbo, API Issues. Share this story by email

Sequel AI Longevity Assistant Open-Sourced, With Personalized Health Insights and User Privacy

Explore more of our coverage of: AI Longevity Assistant, Open-Source Health Data, User Privacy. Share this story by email

Meta Launches Next-Gen MTIA (Meta Training and Inference Accelerator) Chip

Explore more of our coverage of: Meta, AI Infrastructure, MTIA Chip. Share this story by email

Udio Launches Free Beta Music Creation Tool

Explore more of our coverage of: AI Music Generation, DeepMind Researchers, Beta Software Launch. Share this story by email

That’s it! More AI news tomorrow!