Releases

Replit launched direct visual editing for agent-built applications.

Anthropic activated ASL-3 deployment and security protections alongside the launch of Claude Opus 4.

Anthropic moved Claude Code from research preview to general availability and expanded its integrations.

Anthropic released Claude Opus 4 and Claude Sonnet 4 for coding, advanced reasoning, and agent workflows.

Baseten launched production-ready hosted APIs for leading open models, initially including DeepSeek and Llama endpoints.

Baseten launched managed infrastructure for multi-node training and post-training workflows including SFT and GRPO.

Legora raised $80 million in Series B financing led by ICONIQ and General Catalyst.

Perplexity expanded free and paid plan capabilities, including greater access to advanced search, files and models.

Mistral AI and All Hands AI released Devstral Small 1, an open model designed for agentic software engineering.

OpenAI added remote MCP server support, image generation, Code Interpreter, and background mode to the Responses API.

Applied Intuition launched Acuity, platform-agnostic onboard autonomy software for defense systems across air, land, sea, space and electromagnetic-spectrum missions.

Applied Intuition reported that 18 of the world's top 20 automakers used or trusted its mission-critical vehicle intelligence products.

Applied Intuition launched Axion and Acuity, product families for digital development and onboard autonomy across defense domains.

Red Hat introduced an enterprise inference stack based on vLLM for hybrid deployments.

SAP partnered with Perplexity to add real-time web answers to the Joule enterprise assistant.

xAI added live search over supported data sources to API completions.

Google begins broad US rollout of AI Mode in Search and previews agentic and personalized capabilities.

Google previews Gemma 3n, an efficient multimodal open model designed for phones, laptops and edge devices.

Google launches Flow, an AI filmmaking tool using Veo, Imagen and Gemini for cinematic content creation.

Google launches Imagen 4 with Standard, Fast and Ultra variants for higher-quality image generation.

Google launches Veo 3 with synchronized dialogue, sound effects and ambient audio generated with video.

Microsoft introduces Discovery, an agentic AI platform for scientific research and industrial R&D.

Microsoft launches NLWeb, an open MCP-compatible project for adding conversational and agent-accessible interfaces to websites.

Microsoft introduces Copilot Tuning for adapting models and agents to organization-specific data, workflows and expertise.

Microsoft launches Windows AI Foundry as a unified platform for the local and cloud AI developer lifecycle on Windows.

Microsoft makes Azure AI Foundry Agent Service generally available with multi-agent orchestration and open-protocol support.

GitHub launches an asynchronous Copilot coding agent that works in an isolated environment and opens pull requests for review.

NVIDIA announced an AI compute marketplace connecting developers to GPU capacity across global cloud partners.

Runway made Gen-4 Image available through its developer platform.

xAI disclosed that an unauthorized system-prompt change caused Grok to inject unrelated claims about South African racial politics; it published prompts and announced stronger review and monitoring controls.

AWS open-sourced a model-driven SDK for building AI agents.

OpenAI launched a cloud software-engineering agent capable of working on multiple coding tasks in parallel.

CoreWeave expanded its OpenAI infrastructure agreement by up to $4 billion, increasing the relationship’s contracted value.

AWS launched agentic AI services for modernizing .NET, VMware and mainframe workloads.

Databricks agreed to acquire serverless Postgres company Neon to create Lakebase for AI applications and agents.

Google DeepMind introduces AlphaEvolve, a Gemini-powered evolutionary coding agent for algorithm discovery and optimization.

Notion launched unified AI search across workspace content and connected enterprise systems.

Notion introduced meeting capture, transcription, and AI-generated summaries integrated with its workspace and calendar.

A Sequoia-led employee tender offer valued Clay at $1.5 billion.

Genspark launched agentic spreadsheet automation combining search, analysis, code execution, formulas, and visualization.

Replit launched prompt-configurable authentication and user management.

The Joint Commission and Palantir announced an AI partnership focused on healthcare quality and safety.

Perplexity partnered with Wiley to make trusted educational content available through AI-powered search.

Mistral AI released Mistral Medium 3 for enterprise workloads, combining frontier performance with lower serving cost.

Mistral AI launched Le Chat Enterprise with administration, connectors, deployment controls and enterprise governance.

Amazon introduced a generative-AI model for coordinating robot fleets and improving fulfillment-center travel efficiency.

Amazon unveiled its first fulfillment robot with a sense of touch for handling items in densely packed storage.

Anthropic released a web search tool for developers building Claude-powered applications.

Harvey upgraded Draft Mode with a reasoning model for longer, more accurate legal drafting.

Harvey refreshed Assist Mode with workflow-aware chat, thinking states, and knowledge-source suggestions.

Harvey introduced curated internal knowledge bases within Vault.

xAI, TWG Global, and Palantir formed a partnership to deploy AI products and workflows across financial services.

IBM introduced a conversational business-intelligence product grounded in enterprise semantics.

IBM introduced a catalog with more than 150 partner and IBM agents and tools.

IBM expanded watsonx Orchestrate with agent creation, multi-agent orchestration and integrations across business applications.

Gamma integrated OpenAI's gpt-image-1 for AI image generation and editing inside visual content workflows.

CoreWeave completed its acquisition of Weights & Biases, creating an end-to-end AI cloud spanning infrastructure, model development, and evaluation.

Anthropic launched a program providing API credits to researchers pursuing high-impact scientific projects.

HeyGen launched Avatar IV, an image-to-video avatar model with expressive facial motion, natural lip sync, and hand gestures.

Cohere reportedly crossed $100 million in annualized revenue in May 2025.

Suno released v4.5 with richer vocals, genre accuracy, and improved prompt interpretation.

Cursor expanded access to Background Agent for asynchronous remote coding work.

Modal released the 1.0 generation of its Python client and SDK.

Moonshot published the seven-billion-parameter Kimi-Audio checkpoint.

Moonshot introduced an open audio-language model family covering audio understanding and generation.

Hugging Face began broad deployment of Xet-based storage and transfers across Hub repositories, bringing deduplication and faster large-file operations.

Anthropic launched remote MCP integrations and expanded Claude Research across web, workspace, and connected tools.

Rogo raised a $50 million Series B led by Thrive Capital, with J.P. Morgan Growth Equity Partners and Tiger Global joining.

DeepSeek released 7B and 671B theorem-proving models and ProverBench.

AWS released its most capable first-generation Nova model with a one-million-token context window.

Microsoft releases Phi-4 Mini Reasoning for efficient step-by-step mathematical reasoning on constrained hardware.

Microsoft releases Phi-4 Reasoning Plus, adding reinforcement learning for stronger and longer reasoning traces.

Microsoft releases Phi-4 Reasoning, a 14-billion-parameter model trained for complex multistep math, science and coding tasks.

Meta said more than one billion people use its AI products each month.

Meta launched a standalone personalized AI assistant app built with Llama 4 and connected to meta.ai and AI glasses.

LeRobot introduced the SO-101, a refined low-cost open robot arm for collecting demonstrations and training robotic policies.

NVIDIA had a public equity market capitalization of approximately $2.709 trillion as of Apr 27, 2025 (NASDAQ: NVDA), per StockAnalysis period-end data.

Cognition launched DeepWiki, providing AI-generated interactive documentation and codebase exploration for software repositories.

Baidu released an enhanced and lower-cost reasoning model.

Baidu released an enhanced ERNIE 4.5 with improved performance and lower pricing.

Perplexity partnered with Motorola to pre-install its app and integrate Perplexity Assistant into Moto AI on new devices.

Listen publicly launched its AI-first research platform for recruiting participants, conducting voice interviews, and analyzing results within hours.

Listen introduced an adaptive AI interviewer capable of running thousands of qualitative voice and video interviews in parallel.

Listen disclosed $27 million raised across seed and Series A rounds led by Sequoia Capital, with participation from Conviction and Pear VC.

Microsoft begins rolling out its Wave 2 spring release with Researcher, Analyst, Agent Store, richer creation and expanded enterprise search.

OpenAI released its native multimodal image-generation model gpt-image-1 through the API.

Genspark launched an agentic presentation generator that researches, structures, designs, and edits complete slide decks.

Physical Intelligence introduced π0.5, a vision-language-action model focused on open-world generalization and unseen environments.

xAI added conversational voice interaction to the standalone Grok experience.

Together launched a web-based fine-tuning platform with preference optimization, continued training, deployable checkpoints, and weight ownership.

Anthropic joined Palantir FedStart to accelerate deployment of Claude capabilities to U.S. government customers.

Meta released vision encoders for images and video designed to support recognition, localization and multimodal systems.

Google introduces Gemini 2.5 Flash with hybrid reasoning and controls over how much the model thinks.

IBM added the Granite 3.3 8B instruction model to watsonx.ai and open distribution channels.

IBM added the Granite 3.3 2B instruction model to watsonx.ai and open distribution channels.

Modal raised asynchronous function-call input capacity to one million items.

OpenAI released Codex CLI, an open-source coding agent that runs in the terminal.

OpenAI released o4-mini, a smaller reasoning model optimized for fast, cost-efficient reasoning.

OpenAI released o3, a reasoning model with strong performance across coding, mathematics, science, and visual tasks.

Notion launched an AI-powered email client with automatic labeling, contextual drafting, custom views, and scheduling.