Qwen3.8-Omni-Flash Launches: Multi-Speaker Error Rate Drops to 3%, API Price Cut 98%

Qwen released Qwen3.8-Omni-Flash, boosting average scores by 25% over its predecessor and cutting multi-speaker meeting transcription error rate from 88% to 3%, while slashing audio API prices by 98% and shipping two companion frameworks.

2026-09-18 ~ 2026-09-18 · 2 related posts

• Qwen3.8-Omni-Flash: meeting ASR errors cut from 88% to 3%, API prices down 98% — karminski3 · 2026-09-18

1 near-duplicate retellings: karminski3