News

the-decoder16d
Salesforce's CRM benchmark finds AI agents struggle in real-world business scenarios
CRMArena-Pro is designed to test how well large language models (LLMs) can function as agents in real-world business settings, especially for CRM tasks like sales, customer service, and pricing. The ...
the-decoder17d
Anthropic shares blueprint for Claude Research agent using multiple AI agents in parallel
Anthropic has shared the design for its new research agent, which uses a multi-agent approach: a main agent analyzes questions, creates strategies, and assigns specialized sub-agents to work on ...
the-decoder17d
OpenAI updates ChatGPT search with smarter answers and image search
OpenAI has rolled out a major update to ChatGPT's integrated search, introducing smarter answers, better handling of long conversations, and a new image search feature. According to OpenAI, the ...
the-decoder2mon
Safety assessments show that OpenAI's o3 is probably the company's riskiest AI model to date
OpenAI’s latest language models, o3 and o4-mini, incorporate advanced reasoning capabilities and extensive tool use, including image analysis, Python execution, and web browsing.According to OpenAI, ...
the-decoder2mon
Meta releases first multimodal Llama-4 models, leaves EU out in the cold
EU excluded from using Llama 4 multimodal models. Meta is releasing the Llama 4 models under its standard Llama license, but with a new limitation: companies and individuals based in the EU are ...
the-decoder2mon
Meta Neuroscientist King: "Some of the concepts like reasoning may need to be re-evaluated"
The neuroscientist Jean-Rémi King leads the Brain & AI team in Meta’s AI division. In an interview with The Decoder, he discusses the connection between AI and neuroscience, the challenges of ...
the-decoder1y
OpenAI's new multimodal "GPT-4 omni" combines text, vision, and audio in a single model - THE DECODER
OpenAI has unveiled GPT-4o, its latest large multimodal model that sets new standards in performance and efficiency by combining text, image, and audio processing in a single neural network. The "o" ...
the-decoder1mon
Anthropic releases Claude 4 with new safety measures targeting CBRN misuse
Anthropic has released its next generation of AI models, Claude Opus 4 and Claude Sonnet 4, and is introducing new safety measures designed to prevent their use in developing chemical, biological, ...
the-decoder1mon
OpenAI brings its new GPT-4.1 model to ChatGPT users
OpenAI is rolling out its GPT-4.1 model to ChatGPT, making it available outside the API for the first time. Plus, Pro, and Team users can now select GPT-4.1 from the "More models" drop-down menu, ...
the-decoder1y
Elon Musk's xAI plans to build a "Gigafactory of Compute" to dwarf Meta's massive GPU clusters - THE DECODER
Elon Musk plans to build an AI supercomputer with Nvidia graphics cards for his AI company xAI, according to a report from The Information, which cites a presentation Musk gave to investors. The ...
the-decoder1mon
Wait a minute! Researchers say AI's "chains of thought" are not signs of human-like reasoning
In a position paper, the authors argue that interpreting these so-called "chains of thought" as evidence of human-like reasoning is both misleading and potentially harmful for AI research.. The team, ...
the-decoder1y
Midjourney gets personalized models, v7 before summer, video by end of year - THE DECODER
During the recent "Office Hours", Midjourney founder David Holz gave a preview of the upcoming features of the popular AI image generator. According to Holz, version 7 is on track and will be released ...