ChatGPT web can actually listen to audio, but the same model can't via API or Codex
Possible_Statement84 · reddit · 2026-10-23
A Reddit user documented a strange asymmetry: ChatGPT web genuinely understands audio, but the same model accepts no audio input via API or Codex.
Experiments: a pure test beep was described correctly with no code executed; a Russian emo-rap track renamed to hide its identity was read via soundfile in Python (26 seconds, display(Audio(...))) — no Whisper or ASR anywhere — after which the model accurately described instruments, structure and lyrical themes, missing exactly the parts the human couldn't make out either. On another track it replayed a chunk at 0.9x speed and got roughly 70% of words right.
The catch: the model itself claims not to support audio, and a tool once returned "you do not support audio" mid-conversation. The help page only says audio understanding depends on the selected model. Suspected: a separate audio path or model living only in the web harness. (model: gpt-6, Extra High)
More from Models
- Dream Labs emerges from stealth with Dream-1 robot foundation model at $3/hour — scott_e_reed · 2026-10-23
- Claude Models on Google Cloud (Anthropic managed) is now generally available — claudeai · 2026-10-23
- Beam vs GLM 5.3 Abliterated: 10-round attack/defense test shows 96% of vulns patched — bigblueboo · 2026-10-23
- MiniMax M3.1 Flash free for 3 days on 0G, with 1M context and 512K output — MiniMax_AI · 2026-10-23
- TLA+ expert corrects Grok: TLC only exhausts finite state spaces, not all behaviors — tianyin_xu · 2026-10-23
- Open LLM cost calculator shows Arabic tokenization and 6x price gaps across models — Moham-saed · 2026-10-23