Releases

OpenAI released gpt-oss-120b and gpt-oss-20b under an open-weight license.

Perplexity integrated OpenTable so users could find and book restaurant reservations from answers.

IBM expanded Granite with open speech-recognition models for enterprise transcription.

Baidu Wenku and Baidu Drive launched a multi-agent productivity platform.

Google releases Deep Think, a parallel-thinking mode based on Gemini 2.5, to Gemini Ultra subscribers.

CoreWeave closed a $2.6 billion secured delayed-draw term loan to finance infrastructure supporting its OpenAI agreement.

Cohere released its first commercial multimodal Command model for enterprise document, chart, OCR and scene understanding.

The launch of HF Jobs established an on-demand compute layer alongside the Hub’s storage, collaboration, and inference products.

Hugging Face launched pay-per-use compute jobs that run scripts and containers on CPUs and GPUs directly through the hf CLI.

Gannett joined Perplexity's publisher program, broadening licensed news participation.

Mistral AI released Codestral 25.08 and expanded its enterprise coding stack across Mistral Code and developer tools.

Mark Zuckerberg outlined Meta's strategy to build personal superintelligence and distribute it through consumer products.

Cohere and Bell partnered to deploy sovereign enterprise AI for telecommunications workloads in Canada.

Oracle added xAI's Grok 4 model to OCI Generative AI.

GitHub opens Spark to public preview for building and deploying full-stack applications from natural-language prompts.

Oracle committed to operate and expand large-scale AI infrastructure for OpenAI's Stargate program.

OpenAI and Oracle agreed to develop 4.5 gigawatts of additional U.S. Stargate data-center capacity.

Perplexity raised approximately $100 million at an $18 billion valuation.

Le Chat added Deep Research, multilingual reasoning, image editing, projects and expanded voice capabilities.

OpenAI launched ChatGPT agent, combining web interaction, research, and execution capabilities in a unified agent.

Scale reduced approximately 200 employee roles and contractor assignments after determining it had expanded its generative-AI capacity too quickly, while prioritizing enterprise, government, and international growth.

AWS introduced managed runtime, memory, identity, gateway, observability, browser and code-execution services for AI agents.

Mistral AI released Voxtral Small and Voxtral Mini, open speech-understanding models for transcription and audio reasoning.

Anthropic launched an integrated financial-analysis solution with data connectors, Claude Code, and enterprise deployment support.

xAI launched its government suite and announced a U.S. Department of Defense agreement with a $200 million ceiling, plus availability through the GSA schedule.

xAI added animated AI companions to the Grok mobile experience.

AWS introduced an agentic AI development environment centered on specifications and automated hooks.

Meta announced Hyperion, a Louisiana AI data-center campus planned to scale to five gigawatts of compute capacity.

Moonshot published the post-trained Kimi K2 Instruct checkpoint for agentic and coding tasks.

Moonshot published the base pretrained Kimi K2 checkpoint.

Moonshot introduced Kimi K2, a one-trillion-parameter mixture-of-experts model with 32B active parameters.

Mistral AI expanded the Devstral family with Devstral Small 1.1 and a larger Devstral Medium model.

Hugging Face and Pollen Robotics launched Reachy Mini, a compact open-source desktop robot aimed at developers, researchers, and educators.

Perplexity launched Comet, an AI-native browser with an embedded assistant, initially for Max subscribers and waitlisted users.

xAI removed inappropriate Grok posts containing antisemitic tropes and praise for Hitler, tightened hate-speech controls and later explained that a system update had made the bot too susceptible to extremist user content.

xAI launched Grok 4 with native tool use and real-time search, plus the higher-compute Grok 4 Heavy multi-agent variant.

Microsoft releases Phi-4 Mini Flash Reasoning for fast reasoning on edge, mobile and real-time systems.

Hugging Face released a fully open 3B multilingual model with 128K context, switchable reasoning, tool use, and a disclosed training recipe.

CoreWeave announced an all-stock agreement to acquire Core Scientific and increase direct ownership of power and HPC data-center infrastructure.

Mistral AI introduced AI for Citizens, a sovereign AI offering for governments and public-sector institutions.

Perplexity launched Max, a premium subscription tier with expanded access to frontier models and new products.

Baidu introduced enterprise AI agents designed as role-specific digital workers.

Morgan Stanley arranged $5 billion of debt financing and xAI separately raised $5 billion in strategic equity to fund AI infrastructure and product development.

Baidu open-sourced ten ERNIE 4.5 variants spanning dense and mixture-of-experts text and vision-language models.

Meta consolidated its frontier AI efforts into Meta Superintelligence Labs, led by Chief AI Officer Alexandr Wang with Nat Friedman overseeing AI products and applied research.

Google releases Gemma 3n weights and tools for efficient multimodal AI on edge devices.

Oracle and xAI made Grok 3 and Grok 3 Mini available through OCI Generative AI.

Google DeepMind introduces AlphaGenome to predict regulatory effects of DNA sequence variants at high resolution.

Google DeepMind introduces an on-device vision-language-action model that runs locally on robots.

Google transfers the A2A protocol project to the Linux Foundation to support neutral, industry-wide governance.