--- schema: 1 type: video title: >- Reachy Mini: the $300 open source robot you can actually hack — Andres Marafioti, Hugging Face description: >- Qwen3-TTS shipped at 0.8x real time: one second of audio took 1.2 seconds to generate. Andres Marafioti from Hugging Face spent two weeks fixing it. The culprits were no streaming, 500 autoregressive steps per audio packet with a CPU GPU round trip on each, and a dynamic KV cache that blocked compilation. Static KV cache plus CUDA graph captures brought it to 5.8x real time with time to first audio under 200 milliseconds. The platform is Reachy Mini, a $300 open source robot Hugging Face has shipped to 7,500 people. It arrives unassembled. Talking to it is their most used app by far. The voice stack runs Parakeet transcription every 150 milliseconds with partial results feeding back to the robot mid-sentence, Qwen 3.5 27B for the LLM, and this optimized TTS. At that speed, infrastructure round trips match model latency, so the load balancer separates LLM endpoints from conversation nodes to handle the difference in how much different users actually talk. --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/0jeZfjJMfmo-reachy-mini-the-300-open-source-robot-you-can-actually-hack-andres-marafioti-hugging-face Join our upcoming AI Engineer conferences (see https://ai.engineer landing page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS 2026 in Sydney --- canonical: 'https://liquidgold.cc/watch/0jeZfjJMfmo' markdown: 'https://liquidgold.cc/watch/0jeZfjJMfmo.md' llms: 'https://liquidgold.cc/watch/0jeZfjJMfmo/llms.txt' parent: 'https://liquidgold.cc/channel/aidotengineer.md' channel: aidotengineer updated: '2026-10-24T10:12:58.035Z' language: null children: [] videoId: 0jeZfjJMfmo durationSeconds: 1276 publishedAt: '2026-05-29T17:00:06.000Z' searchChunks: - ordinal: 0 text: null startSeconds: 26 endSeconds: 159 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=26' heldReason: rights - ordinal: 1 text: null startSeconds: 26 endSeconds: 159 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=26' heldReason: rights - ordinal: 2 text: null startSeconds: 170 endSeconds: 250 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=170' heldReason: rights - ordinal: 3 text: null startSeconds: 260 endSeconds: 381 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=260' heldReason: rights - ordinal: 4 text: null startSeconds: 260 endSeconds: 381 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=260' heldReason: rights - ordinal: 5 text: null startSeconds: 260 endSeconds: 381 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=260' heldReason: rights - ordinal: 6 text: null startSeconds: 381 endSeconds: 392 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=381' heldReason: rights - ordinal: 7 text: null startSeconds: 392 endSeconds: 483 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=392' heldReason: rights - ordinal: 8 text: null startSeconds: 392 endSeconds: 483 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=392' heldReason: rights - ordinal: 9 text: null startSeconds: 587 endSeconds: 676 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=587' heldReason: rights - ordinal: 10 text: null startSeconds: 587 endSeconds: 676 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=587' heldReason: rights - ordinal: 11 text: null startSeconds: 687 endSeconds: 822 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=687' heldReason: rights - ordinal: 12 text: null startSeconds: 687 endSeconds: 822 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=687' heldReason: rights - ordinal: 13 text: null startSeconds: 687 endSeconds: 822 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=687' heldReason: rights - ordinal: 14 text: null startSeconds: 832 endSeconds: 931 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=832' heldReason: rights - ordinal: 15 text: null startSeconds: 832 endSeconds: 931 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=832' heldReason: rights - ordinal: 16 text: null startSeconds: 942 endSeconds: 1077 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=942' heldReason: rights - ordinal: 17 text: null startSeconds: 942 endSeconds: 1077 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=942' heldReason: rights - ordinal: 18 text: null startSeconds: 942 endSeconds: 1077 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=942' heldReason: rights - ordinal: 19 text: null startSeconds: 1088 endSeconds: 1181 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=1088' heldReason: rights - ordinal: 20 text: null startSeconds: 1088 endSeconds: 1181 watch: 'https://liquidgold.cc/watch/0jeZfjJMfmo?t=1088' heldReason: rights --- # Reachy Mini: the $300 open source robot you can actually hack — Andres Marafioti, Hugging Face Qwen3-TTS shipped at 0.8x real time: one second of audio took 1.2 seconds to generate. Andres Marafioti from Hugging Face spent two weeks fixing it. The culprits were no streaming, 500 autoregressive steps per audio packet with a CPU GPU round trip on each, and a dynamic KV cache that blocked compilation. Static KV cache plus CUDA graph captures brought it to 5.8x real time with time to first audio under 200 milliseconds. The platform is Reachy Mini, a $300 open source robot Hugging Face has shipped to 7,500 people. It arrives unassembled. Talking to it is their most used app by far. The voice stack runs Parakeet transcription every 150 milliseconds with partial results feeding back to the robot mid-sentence, Qwen 3.5 27B for the LLM, and this optimized TTS. At that speed, infrastructure round trips match model latency, so the load balancer separates LLM endpoints from conversation nodes to handle the difference in how much different users actually talk. --- Watch along with the corrected transcript, summary, timestamps and resources on this talk's official AIE page: https://ai.engineer/talks/0jeZfjJMfmo-reachy-mini-the-300-open-source-robot-you-can-actually-hack-andres-marafioti-hugging-face Join our upcoming AI Engineer conferences (see https://ai.engineer landing page): - AIEi Shanghai, November 5–6 — the first for meeting top Chinese labs and engineers - AIE CODE in SF, November 10–12 — the top AI Coding conference returns - AIEi Sydney, December 7–8 — the first AIEi run alongside NeurIPS 2026 in Sydney --- Canonical: https://liquidgold.cc/watch/0jeZfjJMfmo · Channel: [AI Engineer](https://liquidgold.cc/channel/aidotengineer.md) > Search chunk text is held until public-source provenance is verified. ## Document chunks > Missing: no stored document chunks are available.