OpenAI Launches GPT Live Transcribe and GPT Transcribe with Up to 41% Lower Error Rates
reach_vb · x · 2026-07-29
OpenAI has released two new transcription models in the API: GPT Live Transcribe for low-latency live transcription and GPT Transcribe for completed audio and batch workloads. Key improvements: Live error rate 9.60% vs previous 11.65% (18% lower), Transcribe error rate 8.98% vs Whisper-1's 15.21% (41% lower). Adding context improves semantic accuracy from 38.5% to 44.6% for Live and 41.6% to 45.2% for Transcribe. Supports free-form context, keywords, and multiple languages, useful for voice agents, calls, captions, and meetings.
More from Multimodal
- Elorian launches Glimpse, an Astra-level visual reasoning model at 17x lower cost — CurieuxExplorer · 2026-10-17
- MOSS-Image base model showcase: cfg4, 39 steps, ~1.5MP renders — COMPLOGICGADH · 2026-10-17
- MOSS image model 4K test: 55s per frame on RTX 5090, 19.5GB peak VRAM — RauschCraft · 2026-10-17
- Fizgig LoRA Isolation: 0/50 face bleed-through vs 48/50, likeness kept — shootthesound · 2026-10-17
- MiniCPM-V 4.7 runs a 35B multimodal model in 3GB of memory, fully on-device — alifcoder · 2026-10-17
- Open-source .char format now supports up to 4 consistent characters per generation — ashishsanu · 2026-10-17