GPT-5.5 Reportedly Dumber Due to Silent Truncation
bdsqlsz · x · 2026-07-05
Developer @bdsqlsz noticed GPT-5.5 getting dumber, likely due to silent truncation. System prompts inject data every 30 seconds, causing the model to auto-truncate tokens after reaching a certain length, with reasoningtokens following a 518n-2 pattern. The author provides a fix: modify the system prompt via config.toml, set disableresponsestorage=true, and point modelinstructionsfile to a custom instruction file.
Related event: GPT-5.5 Reportedly Degrades Due to Silent Token Truncation(3 posts)→
More from Models
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11
- Opus Refuses Protein Research Codebase Over 'Safety' Concerns, Dev Considers Rolling His Own — josephdviviano · 2026-09-11
- User Hails Unconfirmed 'DeepSeek 4.1 Flash' as an Inflection Point in LLMs — himanshustwts · 2026-09-11
- Terminal Bench v4: GLM-5.3 Leads at 41.9%, Kimi-K3 Underwhelms at 12.6% — Ok_Warning2146 · 2026-09-11
- GPT-6 Astra beats Factorio with enemies in 44 in-game hours at ~$4,500 API cost — liminal_bardo · 2026-09-11