Releases

DeepSeek released the experimental V3.2 model with sparse attention.

NVIDIA released GR00T N1.6, Cosmos Reason NIM and new Newton physics integration for Isaac Lab.

Anthropic released an SDK exposing the agent-building infrastructure used by Claude Code.

Anthropic launched a refreshed Claude Code terminal interface, checkpoints, background tasks, subagents, and a native VS Code extension.

Anthropic released Claude Sonnet 4.5 with major gains in coding, computer use, agents, reasoning, and mathematics.

Kimi added the OK Computer agent mode for executing computer-oriented tasks.

Suno launched its generative audio workstation with multitrack editing and stem generation.

A third OpenAI agreement added up to $6.5 billion, bringing total announced contract value with OpenAI to approximately $22.4 billion.

Perplexity launched a developer Search API exposing its real-time web index, retrieval and content extraction.

xAI expanded its government offering through GSA OneGov, making Grok reasoning models available to U.S. federal employees at negotiated pricing.

Meta launched Vibes, a feed in the Meta AI app and meta.ai for creating, remixing and sharing AI-generated short videos.

Google DeepMind introduces Gemini Robotics 1.5 and an embodied-reasoning model for more capable robot agents.

Gemini 2.5 Flash-Lite becomes generally available for high-volume, latency-sensitive applications.

A second close added $100 million from BDC and Nexxus Capital Management, bringing Cohere's valuation to approximately $7 billion.

Cohere expanded work with AMD across model deployment, compute infrastructure and internal use of North.

Suno released v5 with clearer audio, more natural vocals, creative control, and stronger genre understanding.

OpenAI, Oracle, and SoftBank announced five new US Stargate data-center sites, moving planned capacity toward nearly 7 gigawatts.

DeepSeek improved language consistency and its code and search agents.

The US General Services Administration added Meta's Llama models to OneGov to support federal-agency AI adoption.

OpenAI and NVIDIA signed a letter of intent to deploy at least 10 gigawatts of NVIDIA systems, with the first Vera Rubin gigawatt targeted for the second half of 2026.

OpenAI and NVIDIA announced a strategic partnership to deploy at least 10 gigawatts of NVIDIA systems for OpenAI infrastructure.

Scale released MCP Atlas to evaluate how well AI agents combine tools across real Model Context Protocol servers.

Scale expanded the SEAL Leaderboards into agent evaluation, beginning with SWE-Bench Pro and MCP Atlas.

Scale released a harder, contamination-resistant benchmark using real and commercial repositories to measure software agents on complex multi-file engineering tasks.

xAI released Grok 4 Fast with a 2M-token context window, unified reasoning modes and lower-cost agentic search.