AI moves deeper into products, agents and infrastructure in July 2026

July 2026 brought a broad set of AI releases aimed at making the technology more useful in consumer products and business workflows. Google introduced a smart speaker built around Gemini for Home, with pre-orders beginning at $99.99 and a June 25 launch, and released Gemini 3.5 Live Translate for real-time speech-to-speech translation across 70+ languages. OpenAI previewed GPT-5.6 Sol, Terra, and Luna, while also announcing the “Jalapeño” inference chip with Broadcom.

Anthropic expanded its Claude lineup with Claude Fable 5, Claude Mythos 5 and Claude Sonnet 5, emphasizing stronger reasoning, coding and multimodal capabilities at lower cost. OpenAI reported fivefold growth in active Codex users in H1 2026, a tenfold jump in long-running (> 8-hour) agent tasks, and 13-50x year‑over‑year output gains per developer. Meta’s Brain2QWERTY v2 decoded spoken sentences from MEG scans after training on 22,000 sentences, reaching a 39% word-error rate on a 128-word vocabulary.

Safety and infrastructure also received attention. Google DeepMind detailed an AI Control Roadmap for securing autonomous agents, while AWS added Bedrock managed knowledge bases, web search for agents, WAF bot monetisation, AgentCore Harness and S3 Object Annotations, which support up to 1 GB of structured metadata per object. OpenAI added ChatGPT Enterprise analytics and spending controls, reflecting growing demand for governable AI at scale.