Meta Releases Llama 5 with Multimodal Capabilities · Perplexity AI Launches Enterprise Knowledge Assistant · Ollama Adds

Published: 2026-04-20 18:00 (Asia/Shanghai) Coverage: 2026-04-20 06:00 — 2026-04-20 18:00

📰 Top Stories

1. Meta Releases Llama 5 with Multimodal Capabilities

Meta open-sources Llama 5, featuring native multimodal processing for text, images, audio, and video in a single unified model with competitive performance against proprietary alternatives.

2. Perplexity AI Launches Enterprise Knowledge Assistant

Perplexity AI introduces an enterprise-grade knowledge assistant that connects to internal documents, databases, and APIs to provide searchable, cited answers for corporate knowledge workers.

3. Ollama Adds Enterprise-Grade Model Management

Ollama introduces team and enterprise features including model versioning, access controls, and centralized deployment, making local AI model management viable for large organizations.

4. Microsoft Autogen Adds Human-in-the-Loop Agent Coordination

Microsoft’s AutoGen framework introduces human-in-the-loop coordination, allowing human operators to review, approve, or redirect AI agent actions during multi-agent collaborations.

5. OpenAI Establishes New AI Safety Research Lab in Geneva

OpenAI opens a dedicated AI safety research facility in Geneva, focusing on alignment research, red-teaming, and international AI governance collaboration.

6. Anthropic Claude 4 Achieves Top Scores on Coding Benchmarks

Anthropic’s Claude 4 series achieves state-of-the-art results on SWE-bench and HumanEval benchmarks, outperforming GPT-4.5 in complex code generation and debugging tasks.

7. MIT Develops AI System That Learns from 10 Examples

MIT researchers demonstrate a new few-shot learning architecture that achieves competitive performance with as few as 10 training examples, dramatically reducing data requirements.

📊 Trend Watch

🔮 What to Watch

• Enterprise Adoption: Growing wave of enterprise AI deployments shifting from pilot programs to production-scale rollouts
• Regulatory Landscape: Evolving AI governance frameworks across multiple jurisdictions shaping industry standards

Briefing generated: 2026-04-20 18:00 (Asia/Shanghai) Data sources: AI-curated from public technology reports and industry analysis