AI News Summary 2026-07-03

AI News Summary July 3, 2026

GAFAM and Major AI Companies

Introducing GeneBench-Pro

On June 30, 2026, OpenAI released GeneBench-Pro, a research benchmark designed to measure whether models can handle judgment-based analysis in real-world computational biology. The proposal expands on GeneBench and targets more challenging tasks in genomics, quantitative biology, and translational medicine.

Claude Sonnet 5

On June 30, 2026, Anthropic launched Claude Sonnet 5 and describes it as its most agentic Sonnet model to date, with improvements in coding and tool use, as well as a lower price than Opus.

Influencers and Tech Blogs

Release: llm-coding-agent 0.1a0

On July 2, 2026, Simon Willison published Release: llm-coding-agent 0.1a0, in which he explains how he built a new Python library on top of llm to create a Claude Code-style coding agent with tools for reading and editing files and executing commands.

Generative Imaging

Magnific Plugins

On July 1, 2026, Magnific announced Magnific Plugins, bringing its creative suite directly to After Effects, Premiere Pro, DaVinci Resolve, Final Cut Pro, and Photoshop to reduce the need to switch between tools.

Chatbots and Agents

Hermes Agent v0.18.0 (v2026.7.1)

Hermes Agent released on July 1, 2026 v0.18.0, its “Judgment Release,” featuring zero open P0/P1 issues, a “mixture-of-agents” model with first-class citizens, evidence-based verification, and goal completion contracts.

openclaw 2026.7.1-beta.1

OpenClaw released a beta on July 1, 2026, featuring GPT-5.6 support, external harness attachment, and Telegram/Codex workflows—a useful combination for agent automation and orchestration.

Local AI and Serving

Experience and Lessons Learned from Serving Multi-Stage Qwen3-Omni in vLLM-Omni

vLLM published on July 1, 2026 Experience and Lessons Learned from Serving Multi-Stage Qwen3-Omni in vLLM-Omni, explaining a three-stage pipeline, OpenAI API-compatible serving, and optimizations for throughput and latency.

Faster Gemma 4 on MLX with Multi-Token Prediction

Ollama published Faster Gemma 4 on MLX with multi-token prediction on June 29, 2026, reporting a nearly 90% improvement on Apple Silicon for a coding agent benchmark.