GLM-5.3 API hits frontier range; OpenAI pauses largest frontier RL run for safety
快鲤鱼 · wechat · 2026-08-19
Daily digest highlights: Zhipu's GLM-5.3 API goes live, scoring 60 on the Artificial Analysis Intelligence Index — frontier territory, tied with Kimi K3 as top open model, weights open-sourcing next Friday. OpenAI has paused its largest planned frontier RL training run for two weeks to strengthen safety monitoring, per co-founder Greg Brockman. Korea's Motif trained Motif3 — on par with Qwen3.7 Max — with $25M of state-backed Nvidia B200s, exploring a token-as-a-service model. Also: Cerebras unveiled CS-4 claiming up to 30x GPU inference speed (4,400+ tokens/s/user on GPT-OSS-120B), shipping Q3; multiple US states are tightening data center permitting; Google says AI is automating part of its forward-deployed engineers' work on enterprise data governance.
Related event: Zhipu Launches GLM-5.3 API with 50% Coding Boost(4 posts)→
More from Infra
- Chip equipment beats bull case: 2026 WFE seen at $153B, $278B by 2030 — BenBajarin · 2026-10-16
- Steam now launches in 5.23s, beating Epic's 5.47s — and vLLM gets roasted for slow startup — AlpinDale · 2026-10-16
- Shadeform to Announce Six New Features on Oct 19: Lower GPU Costs, More Suppliers — _jaydeepkarale · 2026-10-16
- Fastino launches Training API for SFT/RL, claiming 25-85% cost cut vs Tinker — kalyan_kpl · 2026-10-16
- Dallas Fed: AI data center debt issuance equals ~1/8 of Treasury duration supply — _onionesque · 2026-10-16
- Running Wan2.1 14B FP16 on a 12GB laptop GPU: VRAM swap stress test finishes in 4m11s — RYuu-1120 · 2026-10-16