-
6 days ago | 
the-decoder.com | Maximilian Schreiner
Summary ByteDance introduces Seedream 3.0, a new text-to-image model. Benchmarks suggest improvements over GPT-4o and Midjourney in speed, accuracy, and visual quality. ByteDance has released Seedream 3.0, a new text-to-image generation model that, according to internal and external evaluations, outperforms its predecessor Seedream 2.0 and rivals or exceeds the quality of current systems like GPT-4o, Midjourney v6.1, and Imagen 3.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
xAI has introduced a new memory feature for its Grok chatbot, allowing it to recall previous conversations and deliver more personalized responses for frequent users. The feature is currently in beta on grok.com and in the iOS and Android apps, but it is not available to users in the European Union or the United Kingdom. Users can delete individual memories directly within the chat interface, and the entire feature can be disabled through the privacy settings.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
Summary The AI startup Perplexity AI is pursuing integration deals with major smartphone manufacturers, securing an agreement with Motorola while continuing discussions with Samsung Electronics despite the latter's strong ties to Google. According to Bloomberg sources familiar with the matter, Perplexity AI is in early talks with Samsung about potential integration options.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
DocuSign plans to release AI-powered contract agents designed to analyze agreements, flag potential risks, and identify issues typically requiring human review. The U.S.-based company, which specializes in electronic signatures and digital contract management, offers software for online contract signing, management, and storage. The new AI agents will be integrated into DocuSign's Intelligent Agreement Management (IAM) platform, initially supporting procurement and sales processes.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
OpenAI is developing a social network that multiple sources familiar with the matter say will resemble X. The project remains in an early stage, with an internal prototype focusing on ChatGPT's image generation capabilities and featuring a social feed. According to a report from The Verge, CEO Sam Altman has already sought feedback from outsiders on the concept. It remains unclear whether the network will launch as a standalone app or integrate as part of ChatGPT.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
Microsoft has launched "Computer Use" for Copilot Studio as an early research preview. This feature enables AI agents to interact with graphical user interfaces and is based on OpenAI's CUA, which also powers Operator. Unlike OpenAI's offering, Microsoft's agents can control not just websites but also desktop applications, and they run entirely on Microsoft's infrastructure. According to Microsoft, company data remains within the Microsoft Cloud and isn't used for model training.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
Summary Anthropic has enhanced its AI assistant Claude with two new capabilities: an agent-based research function and Google Workspace integration, both designed to significantly expand the chatbot's research capabilities. According to Anthropic, the research function goes well beyond the existing web search feature. Claude conducts multiple sequential searches and independently determines what to investigate next.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
Summary OpenAI has formed an internal advisory commission to oversee its nonprofit initiatives, a move that comes as the company faces continued criticism over its shift toward commercial operations. The commission is intended to provide strategic guidance for OpenAI’s philanthropic activities and includes individuals with backgrounds in civil rights, public health, education, and politics.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
Summary Google is rolling out new AI-powered video generation features in its Gemini app and the experimental tool Whisk. Both are based on Veo 2, a video model that, according to Google, is capable of producing realistic motion and detailed visuals. The new functionality is available exclusively to subscribers of the Google One AI Premium plan. In Gemini Advanced, users can now convert text prompts into short video clips.
-
1 week ago | 
the-decoder.com | Maximilian Schreiner
Cohere has introduced Embed 4, a multimodal language model designed for semantic search across complex enterprise documents. The model can process a wide range of content types—including text, images, tables, charts, code, and handwritten scans—commonly found in financial reports, medical records, and industrial documentation. Embed 4 supports files up to 128,000 tokens, or approximately 200 pages, and is compatible with over 100 languages, including Arabic, French, and Japanese.
