Reddit user tallies OpenAI pain: limits halved, $2 burned per $1 earned
Capable_Economist280 · reddit · 2026-09-18
A long Reddit post aggregating user complaints across subreddits: the "Sol" model once handled a full week of intensive work, with one week burning 13 billion tokens; a €50 credit purchase evaporated in minutes. The newer "Astra" model initially offered 2.5x token cost at less than half the time, but server overload forced quality downgrades — "20x" plans dropped from 5 days to 2 days of allowance, and the author's "5x" plan fell from 100% to 25% after 3 hours of work. The author cites the $2 burned per $1 earned figure and OpenAI's deflecting responses, concludes that driving users away may be the only path forward, and claims he'd pay €1,500/month. Model and plan names don't fully match public information, so authenticity is uncertain.
More from Models
- Qwen 3.8 Omni Flash Surfaces with Continuous Video/Audio Understanding and Custom Harness — Mr_Moonsilver · 2026-09-18
- Encoders and decoders are the same thing, and decoders have been doing classification for years — HanchungLee · 2026-09-18
- Sakana AI Launches Fugu Max, a Multi-Agent Orchestrator Routing Tasks to Leanest Capable Models — tkasasagi · 2026-09-18
- Prediction: Every Future LLM Will Ship With a Native 'Jev Mode' — multimodalart · 2026-09-18
- Why don't LLM labs compete on personality? Users value it over raw capability — dioscuri · 2026-09-18
- Fine-tuned 4B model as a decision scorer with temperature-scaled confidence — Gradio · 2026-09-18