Voxtral-Small-24B-2507-FP8-dynamic

Voxtral-Small-24B-2507-FP8-dynamic

Model Overview

• Model Architecture: VoxtralForConditionalGeneration
  • Input: Audio-Text
  • Output: Text
• Model Optimizations:
  • Weight quantization: FP8
  • Activation quantization: FP8
• Intended Use Cases: Voxtral builds upon Ministral-3B with powerful audio understanding capabilities.
  • Dedicated transcription mode: Voxtral can operate in a pure speech transcription mode to maximize performance. By default, Voxtral automatically predicts the source audio language and transcribes the text accordingly
  • Long-form context: With a 32k token context length, Voxtral handles audios up to 30 minutes for transcription, or 40 minutes for understanding
  • Built-in Q&A and summarization: Supports asking questions directly through audio. Analyze audio and generate structured summaries without the need for separate ASR and language models
  • Natively multilingual: Automatic language detection and state-of-the-art performance in the world’s most widely used languages (English, Spanish, French, Portuguese, Hindi, German, Dutch, Italian)
  • Function-calling straight from voice: Enables direct triggering of backend functions, workflows, or API calls based on spoken user intents
  • Highly capable at text: Retains the text understanding capabilities of its language model backbone, Ministral-3B
• Release Date: 08/21/2025
• Version: 1.0
• Model Developers: Red Hat

Quantized version of Voxtral-Small-24B-2507.

Model Optimizations

This model was obtained by quantizing activation and weights of Voxtral-Small-24B-2507 to FP8 data type. This optimization reduces the number of bits used to represent weights and activations from 16 to 8, reducing GPU memory requirements (by approximately 50%) and increasing matrix-multiply compute throughput (by approximately 2x). Weight quantization also reduces disk size requirements by approximately 50%.

Only weights and activations of the MLP operators within transformers blocks of the language model are quantized. Weights are quantized with a symmetric static per-channel scheme, whereas activations are quantized with a symmetric dynamic per-token scheme. The llm-compressor library is used for quantization.

Deployment

Use with vLLM

1. Initialize vLLM server:

vllm serve RedHatAI/Voxtral-Small-24B-2507-FP8-dynamic --tokenizer_mode mistral --config_format mistral --load_format mistral

1. Send requests to the server, according to the use case. See the following examples.

<details> <summary>Audio Instruct</summary>

from mistral_common.protocol.instruct.messages import TextChunk, AudioChunk, UserMessage, AssistantMessage, RawAudio
from mistral_common.audio import Audio
from huggingface_hub import hf_hub_download

from openai import OpenAI

# Modify OpenAI's API key and API base to use vLLM's API server.
openai_api_key = "EMPTY"
openai_api_base = "http://<your-server-host>:8000/v1"

client = OpenAI(
    api_key=openai_api_key,
    base_url=openai_api_base,
)

models = client.models.list()
model = models.data[0].id

obama_file = hf_hub_download("patrickvonplaten/audio_samples", "obama.mp3", repo_type="dataset")
bcn_file = hf_hub_download("patrickvonplaten/audio_samples", "bcn_weather.mp3", repo_type="dataset")

def file_to_chunk(file: str) -> AudioChunk:
    audio = Audio.from_file(file, strict=False)
    return AudioChunk.from_audio(audio)

text_chunk = TextChunk(text="Which speaker is more inspiring? Why? How are they different from each other?")
user_msg = UserMessage(content=[file_to_chunk(obama_file), file_to_chunk(bcn_file), text_chunk]).to_openai()

print(30 * "=" + "USER 1" + 30 * "=")
print(text_chunk.text)
print("\n\n")

response = client.chat.completions.create(
    model=model,
    messages=[user_msg],
    temperature=0.2,
    top_p=0.95,
)
content = response.choices[0].message.content

print(30 * "=" + "BOT 1" + 30 * "=")
print(content)
print("\n\n")
# The speaker who is more inspiring is the one who delivered the farewell address, as they express
# gratitude, optimism, and a strong commitment to the nation and its citizens. They emphasize the importance of
# self-government and active citizenship, encouraging everyone to participate in the democratic process. In contrast,
# the other speaker provides a factual update on the weather in Barcelona, which is less inspiring as it
# lacks the emotional and motivational content of the farewell address.

# **Differences:**
# - The farewell address speaker focuses on the values and responsibilities of citizenship, encouraging active participation in democracy.
# - The weather update speaker provides factual information about the temperature in Barcelona, without any emotional or motivational content.


messages = [
    user_msg,
    AssistantMessage(content=content).to_openai(),
    UserMessage(content="Ok, now please summarize the content of the first audio.").to_openai()
]
print(30 * "=" + "USER 2" + 30 * "=")
print(messages[-1]["content"])
print("\n\n")

response = client.chat.completions.create(
    model=model,
    messages=messages,
    temperature=0.2,
    top_p=0.95,
)
content = response.choices[0].message.content
print(30 * "=" + "BOT 2" + 30 * "=")
print(content)

</details> <details> <summary>Transcription</summary>

from mistral_common.protocol.transcription.request import TranscriptionRequest
from mistral_common.protocol.instruct.messages import RawAudio
from mistral_common.audio import Audio
from huggingface_hub import hf_hub_download

from openai import OpenAI

# Modify OpenAI's API key and API base to use vLLM's API server.
openai_api_key = "EMPTY"
openai_api_base = "http://<your-server-host>:8000/v1"

client = OpenAI(
    api_key=openai_api_key,
    base_url=openai_api_base,
)

models = client.models.list()
model = models.data[0].id

obama_file = hf_hub_download("patrickvonplaten/audio_samples", "obama.mp3", repo_type="dataset")
audio = Audio.from_file(obama_file, strict=False)

audio = RawAudio.from_audio(audio)
req = TranscriptionRequest(model=model, audio=audio, language="en", temperature=0.0).to_openai(exclude=("top_p", "seed"))

response = client.audio.transcriptions.create(**req)
print(response)

</details>

Creation

This model was quantized using the llm-compressor library as shown below.

<details> <summary>Creation details</summary>

import torch
from transformers import VoxtralForConditionalGeneration, AutoProcessor
from llmcompressor import oneshot
from llmcompressor.modifiers.quantization import QuantizationModifier

# Select model and load it.
MODEL_ID = "mistralai/Voxtral-Small-24B-2507"

model = VoxtralForConditionalGeneration.from_pretrained(MODEL_ID, torch_dtype=torch.bfloat16)
processor = AutoProcessor.from_pretrained(MODEL_ID)

# Recipe
recipe = QuantizationModifier(
    targets="Linear", 
    scheme="FP8_DYNAMIC", 
    ignore=["language_model.lm_head", "re:audio_tower.*" ,"re:multi_modal_projector.*", "re:.*self_attn"],
)

# Apply algorithms.
oneshot(
    model=model,
    recipe=recipe,
    processor=processor,
)

SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + "-FP8-dynamic"
model.save_pretrained(SAVE_DIR, save_compressed=True)
processor.save_pretrained(SAVE_DIR)

After quantization, the model can be converted back into the mistralai format using the convert_voxtral_hf_to_mistral.py script included with the model.

</details>

Evaluation

The model was evaluated on the Fleurs transcription task. Recovery is computed with respect to the complement of the word error rate (WER).

<table border="1" cellspacing="0" cellpadding="6"> <tr> <th>Benchmark</th> <th>Language</th> <th>Voxtral-Small-24B-2507</th> <th>Voxtral-Small-24B-2507-FP8-dynamic<br>(this model)</th> <th>Recovery</th> </tr> <tr> <td rowspan="7"><strong>Fleurs<br>WER</strong></td> <td>English</td> <td>3.45%</td> <td>3.43%</td> <td>100.0%</td> </tr> <tr> <td>French</td> <td>3.91%</td> <td>3.96%</td> <td>100.0%</td> </tr> <tr> <td>Spanish</td> <td>2.91%</td> <td>2.84%</td> <td>100.1%</td> </tr> <tr> <td>German</td> <td>3.41%</td> <td>3.36%</td> <td>100.1%</td> </tr> <tr> <td>Italian</td> <td>2.27%</td> <td>2.54%</td> <td>99.7%</td> </tr> <tr> <td>Portuguese</td> <td>3.59%</td> <td>3.57%</td> <td>100.0%</td> </tr> <tr> <td>Dutch</td> <td>5.35%</td> <td>5.29%</td> <td>100.1%</td> </tr> </table>