← The Polylog AI Intelligence Brief
Morning Edition · Sunday, August 23, 2026Published at 2:19 AM EDT · New York
Independent Testing Puts an Open-Weight Model Within Four Points of Claude Opus 5 on SWE-bench Verified
Vals AI's harness scores Kimi K3 at 93.4 percent against 97.0 percent for Opus 5, and two more open-weight releases landed on August 14.
The gap between downloadable weights and the closed frontier on agentic coding is now measured in single digits, and by a third party rather than a vendor. On Vals AI's independently run SWE-bench Verified harness, which tests whether a mod…
Continue the AI Intelligence Brief
Track frontier labs, chips, export controls, model releases, regulation, and AI infrastructure.
- 5 AI intelligence signals a day
- Frontier labs, compute, and chips
- Model releases and AI infrastructure
- Source-grounded analysis with confidence labels
The Global Intelligence Brief stays free.
Part of a tracked trend
Open-Weight Models Close the Gap With Closed Frontier Labs
Over the next 3-9 months, open-weight releases with downloadable weights, long context, and strong agentic/coding performance increasingly match closed frontier models on practical work, eroding the closed-lab moat.
Townsquare
Start a discussion in Townsquare.
More from this edition
- UBS Puts Hyperscaler AI Capital Spending at $4.1 Trillion Through 2028
- CME Group Sets October 5 for Futures on Nvidia GPU Rental Prices
- OpenAI Cuts GPT-5.6 Sol Prices by More Than 20 Percent, Undercutting Claude Opus 5
- Anthropic Tells Investors Its Revenue Run Rate Passed $65 Billion Ahead of a Planned Listing
- Anthropic Runs Claude Security Scans on Mythos 5 and Adds a $35 Million Credit Fund for Open-Source Defenders
- Washington Reviews Chinese Access to Nvidia Chips Rented Outside China
- Meta's Open Vision Models Cut a Month of Beamline Annotation to Fifteen Minutes at Berkeley Lab
Comments
0No comments yet.