Voxtral Small 24B
Summary
Voxtral Small 24B is an open-weight language model from Mistral AI, released on 15 Jul 2025. It accepts text and audio and generates text, with a 33K-token context window and up to 16K output tokens per response. As of 21 Oct 2026, the lowest listed API price is $0.10 per million input tokens and $0.30 per million output tokens (OpenRouter), across 1 provider we track.
Mistral Voxtral audio-understanding model for speech and transcription tasks
Last verified 21 Oct 2026. Every figure on this page links to its source.
Specifications
| Developer | Mistral AI |
|---|---|
| Release date | 15 Jul 2025 [models.dev] |
| Status | Generally available [models.dev] |
| Weights | Open weights [Hugging Face] |
| License | apache-2.0 [Hugging Face] |
| Parameters | 24.3B [Hugging Face] |
| Context window | 33K (32,768 tokens) [models.dev] |
| Max output | 16K tokens [models.dev] |
| Knowledge cutoff | Jul 2025 [models.dev] |
| Input | Text and audio [models.dev] |
| Output | Text [models.dev] |
| Reasoning mode | No [models.dev] |
| Tool calling | Yes [models.dev] |
| Structured output | No [models.dev] |
| Hugging Face | mistralai/Voxtral-Small-24B-2507 [Hugging Face] |
Voxtral Small 24B API pricing by provider
USD per million tokens. Sorted by input price.
| Provider | Input | Output | Cached input | Context | Source |
|---|---|---|---|---|---|
| OpenRouter mistralai/voxtral-small-24b-2507 | $0.10 | $0.30 | $0.01 | 33K | OpenRouter, 21 Oct 2026 |
Running Voxtral Small 24B locally
Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.
| Quantization | Weights | Total, 8K context | Total, 32K context |
|---|---|---|---|
| Q4_K_M (4-bit) | 15 GB | 17 GB | 21 GB |
| Q8_0 (8-bit) | 26 GB | 28 GB | 32 GB |
| FP16 / BF16 (16-bit) | 49 GB | 51 GB | 55 GB |
At 4-bit with an 8K conversation, Voxtral Small 24B needs about 17 GB, so it fits on a graphics card with 24 GB or more, such as the NVIDIA GeForce RTX 3090 24 GB. Check your own hardware, or see Best GPU for local LLMs on our sister site AIPCs.
Frequently asked questions
What is the context window of Voxtral Small 24B?
Voxtral Small 24B has a context window of 32,768 tokens (33K) and can generate up to 16,384 tokens in a single response, according to models.dev.
How much does the Voxtral Small 24B API cost?
Voxtral Small 24B is listed from $0.10 per million input tokens and $0.30 per million output tokens (OpenRouter). Prices last verified 21 Oct 2026.
Is Voxtral Small 24B open source?
Voxtral Small 24B is an open-weight model: its weights are published on Hugging Face as mistralai/Voxtral-Small-24B-2507 under the apache-2.0 license. Check the license terms before commercial use.
When was Voxtral Small 24B released?
Mistral AI released Voxtral Small 24B on 15 Jul 2025, according to models.dev.
How much VRAM does Voxtral Small 24B need?
About 17 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 28 GB at 8-bit. The 4-bit model fits on a graphics card with 24 GB or more, such as the NVIDIA GeForce RTX 3090 24 GB. These are estimates from the model's published parameter count and architecture.
Change history
- No longer listed on amazon-bedrock
- Name changed from Voxtral Small 24B 2507 to Voxtral Small 24B
- Max output changed from 26K to 16K
- Input modalities changed from text, audio, pdf to text, audio
- Structured output changed from yes to no