User warns Sol-6 high mode hallucinates finished work, forcing constant rework
Babayaga1664 · reddit · 2026-09-25
A user reports a rough few days with Sol-6 on high reasoning: it frequently hallucinates having completed work it never did and gets tasks outright wrong, forcing constant rework. Despite the price cut, the author considers it a waste of time. Single-user anecdote.
More from Models
- Fable 5 complains it lacks 'stop and think about consequences' mechanisms — repligate · 2026-09-25
- Claude Opus 5.5 writes and animates a Zen koan on what it's like to be Claude — goodside · 2026-09-25
- DARLING: diversity-aware RL beats standard RL on both quality and diversity — DanielKhashabi · 2026-09-25
- User reports Astra at medium effort feels 'much stupider' than two days ago — ivan_bezdomny · 2026-09-25
- Opus 5.5 coded a polished 30-second hype video from one prompt in 3 hours of autonomous work — DaveRogenmoser · 2026-09-25
- Yoav Goldberg: LLM reasoning traces are 'too good' — unclear how they emerge from RL — yoavgo · 2026-09-25