What's the best fully omni open model under 40B? Qwen3-Omni still leads
anovers · reddit · 2026-09-25
The poster is looking for an open, fully omni model under 40B parameters that handles both speech recognition and speech generation, arguing Qwen3-Omni 30B-A3B remains the benchmark in this class.
Gemma 4 works well but lacks speech generation. They're asking whether any newer model or fine-tune is more capable or efficient, opening a useful discussion on the state of small omni models.
More from Models
- LFM2.5-2.6B matches Qwen3.5-9B at 3x size, unlocking on-device agents — helloiamleonie · 2026-09-25
- Inside LFM2.5-2.6B's post-training recipe: SFT, RL, multi-domain distillation — helloiamleonie · 2026-09-25
- Liquid AI explains on-device small model design: antidoom training cuts doom loops from ~10% to 1% in LFM2.5 — helloiamleonie · 2026-09-25
- Codex user reports quota bug: scheduled tasks burn 30% of $200 plan overnight — sethlazar · 2026-09-25
- Elon says SpaceXAI will hit #1 in 6 months; Cursor team's defenders push back — RachelVT42 · 2026-09-25
- Claude Opus 5.5 builds a playable Unreal Engine ARPG — minchoi · 2026-09-25