69B
- Context
- 33Ktokens
- Released
- 2y ago
- Size
- 46.7Bparameters
Mixtral 8x7B Instruct v0.1 is Mistral AI's instruction-tuned chat model, fine-tuned from the base model Mixtral-8x7B-v0.1 and released under the Apache 2.0 license. It is built for multi-turn instruction following, with its model card showing user and assistant turns formatted in the [INST] chat template.
How it compares
Additional source evidence
- Source model
- mistral/mixtral-8x7b-instruct-v0.1
- Source family
- mistral
- Last updated
- 2023-12-11
- Model context
- 33K
- Model output limit
- 33K
- Open weights
- Yes
- File attachments
- No
- Reasoning
- No
- Tool calling
- No
- Temperature control
- Yes
Downloadable weights
Complete source record (JSON)
{
"id": "mistral/mixtral-8x7b-instruct-v0.1",
"name": "Mixtral 8x7B Instruct v0.1",
"description": "Open sparse mixture-of-experts Mistral instruction model",
"family": "mistral",
"last_updated": "2023-12-11",
"attachment": false,
"reasoning": false,
"tool_call": false,
"temperature": true,
"modalities": {
"input": [
"text"
],
"output": [
"text"
]
},
"limit": {
"context": 32768,
"output": 32768
},
"release_date": "2023-12-11",
"open_weights": true,
"license": "Apache 2.0",
"weights": [
{
"label": "Hugging Face",
"url": "https://huggingface.co/mistralai/Mixtral-8x7B-Instruct-v0.1"
}
]
}Frequently asked questions
Running locally
Estimates from the 46.7B parameter count are 33.1 GB at Q4, 57.6 GB at Q8 and 107.9 GB at FP16. These figures are estimates, covering weights plus about 15% and 0.5 GB for a short context.
No. At Q4, Q8 or FP16 it does not fit on a 24 GB, 16 GB, 12 GB or 8 GB GPU (estimate).
Yes, on large memory sizes. The smallest common Mac memory size that fits Q4 is 48 GB (estimate), assuming macOS lets the GPU use about 75% of unified memory.
Privacy and data
Yes. Mixtral 8x7B Instruct v0.1 has open weights, so you can self-host it for full control of your data.
Capabilities and specs
The context window is 33K tokens (32,768) and the maximum output is 33K tokens (32,768).
No. The input modalities are text only, so it does not read images, audio, video or PDFs.
No. Tool calling (function calling) is not supported.
No. A reasoning or thinking mode is not supported.
Quality and benchmarks
It ranks 354 on Arena Text (Arena's own rank), with a score of 1197.
The closest alternatives, matched on model kind and release date with no comparable price, are Tulu 2 DPO 70B by Ai2, OLMo 1B by Ai2 and Ferret by Apple.
Tulu 2 DPO 70BMixtral 8x7B Instruct v0.1 vs Tulu 2 DPO 70BOLMo 1BMixtral 8x7B Instruct v0.1 vs OLMo 1BFerretMixtral 8x7B Instruct v0.1 vs Ferret
Licensing and access
Mixtral 8x7B Instruct v0.1 has open weights. The license is Apache 2.0.
Yes. The license is Apache 2.0, a permissive license that allows commercial use.
The weights can be downloaded from the Hugging Face repository mistralai/Mixtral-8x7B-Instruct-v0.1.
Basics
Mistral AI made it. It was released on December 11, 2023.
Yes. It has open weights.
Generated from Vertical Tabs data, updated November 11, 2026.
Discussion
No comments yet
Be the first to share your thoughts.