Large Language Models

New Muse Spark model upgrades Meta AI with multimodal input and parallel agents, and is expected to integrate across key Meta platforms such as WhatsApp, Instagram, and Facebook.

OpenAI launches GPT-5.4 with long-context support, improved reasoning, and enterprise-focused variants across ChatGPT, API, and Codex.

Gemini 3.1 Flash-Lite arrives in preview, offering faster throughput for translation, moderation, UI generation, and other large workloads.

MCP is rapidly transforming how AI agents interact with enterprise systems, opening up a new class of supply chain, identity, and governance risks that security teams can’t ignore.

Perplexity’s Model Council runs queries in parallel across multiple frontier models and synthesizes structured insights for enterprise users.

Modal now supports GLM-5 inference and deployment, enabling enterprise teams to test and integrate a large open-weights model via API and custom stacks.