Weekly Recap: Qwen 4, Wan 3.0, Open On-Device TTS, and Apple M6
fromourback · reddit · 2026-08-26
A massive week for open-weight models and hardware, featuring new video, 3D, and audio releases alongside silicon updates.
Models & Video/3D:
- Wan 3.0 & Gen 1.5: Major strides in open generative video coherence and motion.
- Bernini 2 (ByteDance): Open diffusers pipeline for high-detail visuals.
- Qwen 4 / 3.8-Flash-Next: Architectural jumps in efficiency and local inference speed.
- DeepSeek V4 / Flash Vision: Efficient VLM for agentic tooling.
On-Device & Audio:
- Raon-OpenTTS-1B & Audio8-TTS: Ultra-compact TTS for low-VRAM edge devices.
- EchoWM & Evoke: Real-time world-modeling frameworks.
Hardware:
- Apple M6 & M5 Pro: Targeting expanded unified memory for local models.
- PaXini Robot: Advanced tactile sensor integration.
More from Embodied
- Engineer Swaps SystemVerilog for LLM-Generated PALASM Code — satnam6502 · 2026-08-26
- Google Research unveils AgentHands: interactive hand gestures for grounded agents in XR — Google Research · 2026-08-26
- Peking Univ. & MSRA Introduce BCP: Boosting Robot Success Rates via Autonomous Replanning — jiqizhixin · 2026-08-26
- Pemba, the first robot to attempt Everest, launches this fall — 0xAllen_ · 2026-08-26
- Tesla Optimus is secretly preparing for launch — JOBhakdi · 2026-08-26
- Robotics highlights: Dyna-2 scaling laws, Gen1.5 in-context learning — chris_j_paxton · 2026-08-26