haru_vllm
Serve Haru with vLLM-Omni: many concurrent full-duplex conversations on one GPU, over the realtime WebSocket API.
This is a plugin. Installing it registers the Haru model, pipeline and duplex policy with vLLM-Omni; nothing in vLLM-Omni is patched.
Install and run
pip install vllm==0.32.0 vllm-omni==0.32.0
pip install -e ../haru -e . # from this directory
vllm serve higurashi-ai/Haru-4B --omni --deploy-config "$(python -m haru_vllm.deploy)"
The server listens on ws://localhost:8000/v1/realtime?duplex=1. To use it
from the web demo:
python -m haru.server --backend vllm --vllm-url http://localhost:8000
python -m haru_vllm.deploy prints the path of the bundled deploy
configuration (deploy/haru.yaml). Copy and edit
it to change the number of concurrent conversations (duplex_session.max_sessions
and max_num_seqs) or the cache size.
Tested with vLLM 0.32.0 and vLLM-Omni 0.32.0, the releases on PyPI. It also runs
with vLLM 0.33.0 and the vLLM-Omni main branch at commit 04e03a4. The plugin
uses model-runner hooks of vLLM-Omni that may change between versions.
Protocol
The endpoint speaks vLLM-Omni's realtime duplex protocol (OpenAI Realtime events). One conversation:
session.updateopens the session.instructionsis the character card,ref_audioadata:audio/...;base64,URI of the voice. Audio input ispcm_f32leat 24 kHz.- Append one frame (3200 samples) with
input_audio_buffer.append. This first append starts the conversation: the prompt runs and the model's frame 0 comes back. Its audio is not used; send silence. - From then on, every appended frame
kis answered by model framek + 1: oneresponse.output_audio.delta(PCM16, 24 kHz, 3200 samples) and, when the frame has text, aresponse.output_audio_transcript.delta.
Options in session.extra_body, fixed when the session is created:
| Key | Default | Meaning |
|---|---|---|
guidance_scale |
1.5 | Audio guidance, 1 to 2 |
top_k, top_p |
25, 1.0 | Text sampling (with the session's temperature, default 0.7) |
eop_bias |
0 | Added to the logit of the "start speaking" token |
speak_first |
false | The model opens the conversation |
dry |
true | Repetition penalty on the text stream |
seed |
random | Seed of the audio sampling |
How it is built
One autoregressive stage. Each conversation is a single resumable engine
request whose prompt grows by one position per frame; the language model runs
on vLLM's engine (paged attention cache, Gated DeltaNet state, CUDA graphs).
The audio parts run in model hooks of the same step: the user frame is encoded
before the forward pass, and the MeanFlow head and codec decoder run right
after sampling, so a frame's audio leaves in the step that produced it. The
codec runs as Triton kernels replayed from CUDA graphs when triton is
installed (PyTorch's Linux CUDA wheels install it), and in plain PyTorch
otherwise. Set HARU_CODEC_BACKEND=torch (or triton) in the environment of
vllm serve to override that choice.
Metadata
Release files for haru-speech-vllm 0.1.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| haru_speech_vllm-0.1.1.tar.gz | 19.5 kB | With provenance | Details for haru_speech_vllm-0.1.1.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| haru_speech_vllm-0.1.1-py3-none-any.whl | Python 3 | none | any | Details for haru_speech_vllm-0.1.1-py3-none-any.whl |
Total release size: 42.0 kB
Release files / haru_speech_vllm-0.1.1.tar.gz
| Download URL | haru_speech_vllm-0.1.1.tar.gz |
|---|---|
| Size | 19.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
9d0f196b272211f4e0b8b5bdd85fe11d72759516973fba2bfa12064403389bfa
|
|
BLAKE2b-256 checksum How to use checksums |
7a1345831bf6ad26e4e5ecdedca1974edca756692cfce23fc8b27c8b4eddbaee
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 13, 2026.
Transparency logRelease files / haru_speech_vllm-0.1.1-py3-none-any.whl
| Download URL | haru_speech_vllm-0.1.1-py3-none-any.whl |
|---|---|
| Size | 22.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
1319c8a504d5e7596a5ca5fe6a2b14262f8d83d9209d601237e486e74aef4f79
|
|
BLAKE2b-256 checksum How to use checksums |
3eb7af8d7723be17dfa0b0c1458987c345dae01ff5e19676dc8da0cf2d92c31d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 13, 2026.
Transparency logSecurity for haru-speech-vllm 0.1.1
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for haru_speech_vllm-0.1.1.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Nov 13, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.