# AuK / AuK-Flash: Self-Hosting Hardware, Scenarios and Commercial License

> Instruction-driven speech generation, voice cloning, editing, enhancement and source separation where a self-hosted unified audio workflow is more important than a managed API.

- Verified: 2026-09-12
- Released: 2026-09-09
- Canonical: https://chinaapi.ai/open-model-deployment/auk/
- Back to directory: https://chinaapi.ai/open-model-deployment/
- Machine-readable dataset: https://chinaapi.ai/data/open-model-deployment.json

## Model facts

- Vendor: Tencent Hunyuan
- Parameters: 1.5B family model total; Not separately reported active
- Native precision: Each official AuK or AuK-Flash repository totals 6.76 GB: a 6.12 GB BF16 DiT checkpoint plus a shared 637 MB VAE; the external Qwen2.5-Omni-3B encoder is a separate download
- Context: Not applicable
- Category: Speech generation and editing
- Modalities: text, audio → audio
- Serving paths: Official PyTorch CLI and Python API, ComfyUI, SGLang-Omni
- Official weights: AuK base: https://huggingface.co/tencent/AuK?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment; AuK-Flash distilled: https://huggingface.co/tencent/AuK-Flash?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- Model documentation: https://github.com/Tencent-Hunyuan/AuK?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment

## Deployment evidence

- **Official launch / minimum — evidence C:** Tencent documents launching either AuK base or AuK-Flash on cuda:0. It does not name the accelerator or memory floor, and the 6.76 GB model repository excludes the separately downloaded Qwen2.5-Omni-3B encoder plus runtime headroom, so this proves a single-GPU code path rather than a minimum hardware configuration. Source: https://github.com/Tencent-Hunyuan/AuK?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- **200-person team — evidence E:** Start with AuK-Flash and benchmark real clip durations, reference-audio use, request mix and latency. SGLang-Omni documents default maximum batches of 8 for conditioning, 16 for DiT sampling and 4 for equal-length VAE decode, but publishes no capacity result for this 200-seat profile; an operational service also needs a second failure-domain worker.
- **Commercial API — evidence B:** SGLang-Omni provides /v1/audio/speech and /generate routes plus dynamic batching, but caps target duration at 30 seconds by default and does not implement incremental audio streaming. A public service still needs measured capacity, isolation, admission control, storage policy, consent and impersonation-abuse safeguards, and redundant workers. Source: https://sgl-project.github.io/sglang-omni/cookbook/auk.html?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment

## Scenario evidence

- **Zero-shot and described-voice TTS — Vendor-stated:** Tencent documents zero-shot TTS from reference audio and instruction-based TTS from a voice description through the same natural-language interface. Source: https://github.com/Tencent-Hunyuan/AuK?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- **Speech and lyric editing — Vendor-stated:** The official task set covers replacing, inserting or removing spoken content, plus rewriting lyrics while preserving the recording's melody and voice. Source: https://github.com/Tencent-Hunyuan/AuK/blob/main/docs/COOKBOOK.md?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- **Enhancement and source separation — Vendor-stated:** The official recipes cover denoising, dereverberation, speech separation, singing-voice extraction and target-speaker extraction from natural-language instructions. Source: https://github.com/Tencent-Hunyuan/AuK/blob/main/docs/COOKBOOK.md?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- **Lower-latency research with AuK-Flash — Public evidence:** The technical report states that the distilled four-step AuK-Flash achieves a 4.5x wall-clock speedup over AuK under matched conditions. This is a vendor-published experiment, not a ChinaAPI reproduction. Source: https://arxiv.org/abs/2609.08936?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment

## Commercial-use check

- License: MIT — https://github.com/Tencent-Hunyuan/AuK/blob/main/LICENSE?utm_source=chinaapi&utm_medium=research&utm_campaign=open-model-deployment
- MaaS / hosted service: No model-specific MaaS restriction was identified in the published MIT license.
- Attribution: Include Tencent's copyright notice and the MIT permission notice in all copies or substantial portions; the software and weights are provided without warranty.

## Avoid or validate first

- Treating vendor-published quality or speed results as independent or ChinaAPI-reproduced evidence
- Sizing hardware from the 6.76 GB AuK repository alone while omitting the external Qwen2.5-Omni-3B encoder, runtime state and activations
- Exposing voice-cloning or speech-editing features without consent, impersonation-abuse and content-handling controls
- Promising streaming output, high concurrency or an SLA before workload-specific benchmarks

## Known limitations and open questions

- No ChinaAPI reproduction of the downloadable checkpoints, quality claims or 4.5x speed result
- Neither Tencent nor SGLang-Omni publishes an accelerator-memory minimum or throughput result for the defined team and commercial profiles
- The downloadable AuK repository is 6.76 GB, but Qwen2.5-Omni-3B is an additional required encoder download and runtime memory is not included
- SGLang-Omni defaults to a 30-second duration cap and does not provide incremental audio streaming
- Prompt enhancement and optional transcription can introduce separate LLM, ASR, credential and data-handling dependencies

## Evidence boundary

Loading weights, completing a first forward pass and meeting a production latency/SLA are separate thresholds. The 200-person and commercial tiers require workload-specific measurement. License summaries are product research, not legal advice.

Generated by GENERATED BY scripts/gen_open_model_deployment.py.