Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡️ 

Here’s what we got for ya today:

• 🎨 OpenAI’s new image generation is really good

• ✨ Google's Gemini 2.5 drops

• 💬 Microsoft adds ‘deep reasoning’ to Copilot AI

• 🤝 Anthropic Claude + Databricks

Let’s get right into it!

Sign Up | Advertise | Tools

STARTUPS

OpenAI’s new image generation is really good

OpenAI has officially enabled native image generation within GPT-4o for ChatGPT users across all tiers, allowing users to create and refine lifelike images through natural language prompts directly within conversations. Unlike previous models like DALL·E 3, GPT-4o integrates text, code, and images within the same architecture, resulting in more accurate, detailed, and editable visual outputs. The model supports high-fidelity rendering, precise text placement, and consistent styling across interactions, with applications ranging from design and marketing to education and game development.

While limitations remain—such as image cropping, issues with non-Latin text, and detailed edits—OpenAI is addressing these through ongoing improvements. All images are labeled with C2PA metadata for traceability and are subject to content safeguards to prevent misuse.

BIG TECH

Google's Gemini 2.5

Google has unveiled Gemini 2.5, an updated AI model that combines a significantly improved base model with enhanced post-training to deliver better performance across tasks such as coding, mathematics, and reasoning. The new Gemini 2.5 Pro, available via Google AI Studio and for Gemini Advanced subscribers, reportedly outperforms competitors like OpenAI and Anthropic on common AI benchmarks.

Google highlights Gemini’s native multimodality—its ability to understand text, audio, images, video, and code—and teases a forthcoming 2 million token context window for processing larger inputs. CEO Demis Hassabis praised the model’s superior performance in multimodal reasoning and STEM tasks, attributing its quality leap to “reasoning” capabilities built directly into the model, allowing it to break down complex problems and generate more thoughtful, context-aware responses.

BIG TECH

Microsoft adds ‘deep reasoning’ to Copilot AI

Following AI announcements from Google and OpenAI, Microsoft has introduced new features for its Microsoft 365 Copilot, including two advanced “deep reasoning” agents—Researcher and Analyst—as well as enhanced custom AI agent capabilities. Researcher leverages OpenAI’s deep research model to handle complex, multi-step tasks and integrates data from tools like Salesforce and ServiceNow.

Analyst, based on OpenAI’s o3-mini reasoning model, uses chain-of-thought reasoning to analyze raw data, generate spreadsheets, run Python code, and create detailed reports. These tools will begin rolling out in April to early access users, alongside new autonomous agent features in Copilot Studio, which promise to automate a wide range of business workflows using AI-powered, rule-based logic.

STARTUPS

Anthropic Claude + Databricks

Anthropic and Databricks have formed a five-year, $100 million partnership to sell AI tools, focusing on helping businesses build AI agents using Anthropic’s Claude models integrated directly into Databricks’ data platform. The collaboration aims to leverage Databricks’ 10,000+ customer base and Anthropic’s fast-growing presence to target enterprise clients, with both companies’ sales teams promoting each other’s products.

While demand for AI agents is rising, concerns remain about their reliability, which Databricks is working to improve to human-level accuracy. The deal strengthens both companies’ positions in the competitive AI space, where rivals like OpenAI, Google, and Amazon are also active.