AI Daily Brief — 17 June 2026
Simon Willison’s hands-on review crowned GLM 5.2 “probably the most powerful text-only open-weights LLM”, and Vals AI confirmed it as the new open-weight SOTA on the Vals Index, Vibe Code Bench, and Terminal Bench. xAI shipped one-click Grok Build VMs, Cohere posted a digital-sovereignty manifesto, and Midjourney teased its hardware reveal.
Top stories
- Simon Willison: GLM-5.2 is probably the most powerful text-only open-weights LLM. First-person review from a long-time benchmarker carried weight across the community. via Simon Willison
- GLM 5.2 is the new open-weight SOTA on the Vals Index, Vibe Code Bench, and Terminal Bench. Three-benchmark sweep from Vals AI. via Z.ai
- xAI: Use VMs with Grok Build preinstalled with one click. Removes setup friction for SuperGrok / X Premium subscribers. via xAI
- Cohere: digital sovereignty is choice and control over your data. The most direct positioning to date against US-LLM lock-in. via Cohere
- OpenAI’s GPT-5.4 improves a difficult medicinal-chemistry reaction. Brockman highlighted the scientific-use case publicly. via @gdb
Who shipped
AWS added inline-payload support to SageMaker AI Async Inference and shipped autonomous agents in Amazon Quick. Databricks published a payment-fraud-detection walk-through. Midjourney stayed in tease mode ahead of its hardware reveal. Dwarkesh Patel ran a Giordano-Bruno historical aside.
Open-source pulse
GLM 5.2 absorbed the day’s open-weights attention; Zhipu remained the centre of gravity. Cohere’s sovereignty post positioned its stack as the European-aligned alternative.
Quiet corners
Mistral, Meta (Llama), DeepSeek, Anthropic were absent from the cycle.
By the numbers
- 69 stories tracked across 20+ sources
- Most-mentioned model: GLM 5.2
- Most-mentioned lab: Z.ai / Zhipu
- Notable absences: Mistral, Meta, DeepSeek, Anthropic
Compiled by AI Feed’s editor from all 69 headlines published on 17 June 2026.