After days of testing, dev says Sol 6.1 and Astra 6 fail at agentic work, returns to Opus 5.5

NERDDISCO · x · 2026-10-11

After using Sol 6.1 and Astra 6 for several days, NERDDISCO reports both models are poor at agentic work: they ignore explicit user instructions and instead do whatever they think makes sense. He has switched back to Opus 5.5.

Original post →

More from Models

• Zhipu's GLM-5.4/5.5 roadmap leaks: 1T+ params and a full RSI self-training loop — airesearch12 · 2026-10-11
• GPT6.1 Sol fixes weeks-old failing GitHub CI in under 5 minutes — gethackteam · 2026-10-11
• TensorFold 1.0.5 cuts 19.8k-token chat prefill from 8.1s to 0.16s with persistent prompt cache — HankYeomans · 2026-10-11
• Claude Opus 5.5 Takes #1 on Image-to-WebDev Arena, Priced 60% Below GPT-6 Astra — arena · 2026-10-11
• Image-to-WebDev Arena Pareto Frontier: Open Models Hit 92% of Claude at $0.17/M — arena · 2026-10-11
• DeepSeek-V4.1-Flash: bycloud breaks down DeepSeek's boldest architecture overhaul yet — bycloud · 2026-10-11

Models channel →