Can Kimi or GLM replicate recent closed-model math and cyber breakthroughs offline?

Unusual_Guidance2095 · reddit · 2026-07-22

A Reddit post asks whether open models like Kimi K3 or GLM can reproduce the same “world-changing” results recently claimed by closed models.

• The author proposes running models fully offline on local clusters so they cannot browse the web or leak to training data.
• Suggested targets include two recent closed-model math results and some cybersecurity problems, with old GitHub commits used to avoid contaminated solutions.
• The post also asks for verification by qualified mathematicians or security experts, framing the idea as a serious replication test rather than a benchmark score chase.

Original post →

More from Models

• GPT-6 is said to be near, with OpenAI betting on faster inference and custom chips — haider1 · 2026-07-22
• US Open Model Ecosystem Lags Behind China, but Competition Drives Innovation — sudoraohacker · 2026-07-22
• NVIDIA's Ultra Model Handles Massive Codebases and Long Tool Chains — MonaJalal_ · 2026-07-22
• DeepSeek post pairs inference-system optimizations with a 545% theoretical margin analysis — teortaxesTex · 2026-07-22
• Starcloud pitch puts data centers in space back on the AI infrastructure map — ycombinator · 2026-07-22
• Kimi K3’s only real drawback is speed, but the model is otherwise excellent — bindureddy · 2026-07-22

Models channel →