[whatmodel.sh](../../README.md) / Mixtral 8x7B v0.1 (base) # Mixtral 8x7B v0.1 (base) Mistral AI · released 2023-12 · 46.7B params (12.9B active) · apache-2.0 **score 42.7** ±0.9 · #522 of 732 · from 19 benchmark results across 5 sources Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at ● marks Mixtral 8x7B v0.1 (base) among every model on that task or test, worst to best. 
task    vs the others          rank  counts  tests
coding  ▁▁▁▃●▇█▇▅▄▃▁    #400 of 469     51%      7
math    ▁▁▃▅█▆●█▄▂▁▁   #189 of 431*     11%      2
coding · #400 of 469 · 51% of its score Beats [DeepSeek R1 Distill Qwen 7B](deepseek-r1-distill-qwen-7b.md), [Qwen3.5 2B](qwen3-5-2b.md), [DeepSeek R1 Distill Llama 8B](deepseek-r1-distill-llama-8b.md) here, though they rank higher overall. Trails [Qwen2.5 1.5B Instruct](qwen2-5-1-5b.md), [WizardLM 13B V1.2](wizardlm-13b-v1-2.md), [Falcon3 3B Instruct](falcon3-3b.md) here, though they rank lower overall. 
test                                result  vs the others in its run  difficulty        counts  why
HumanEvalSynthesize (HumanEvalPac…   43.3%  ▁·▂▁▂▆▅●▄▄▁▂ beats 22/35  medium · avg 62%     13%  at its level
HumanEval (maker)                    40.2%  █···█·███··● beats 5/5    medium · avg 61%     12%  sharp · at its level
HumanEvalExplain (HumanEvalPack, …   20.7%  ▄▇█▃●▇▄▁▇▃·▁ beats 16/35  medium · avg 46%    8.3%  at its level
HumanEvalFix (HumanEvalPack, 6-la…   14.3%  ▁·▁▁▄█●▄█▁▅▅ beats 14/35  medium · avg 51%    8.0%  sharp · at its level
MBPP (maker)                         60.7%  ▄··▄·▄··█··● beats 5/5    easy · avg 69%      4.2%  at its level
CRUXEval-I                           39.3%  ▁··▁▅▂●▇▁··▁ beats 10/22  medium · avg 52%    2.8%  at its level
CRUXEval-O                           40.5%  ▁·▂▃▁●▁▂···▁ beats 11/22  medium · avg 51%    2.7%  blunt · at its level
math · #189 of 431 · 11% of its score Beats [MiniMax-M2.1](minimax-m2-1.md), [DeepSeek V3 0324](deepseek-v3-0324.md), [DeepSeek V3 (Dec '24)](deepseek-v3.md) here, though they rank higher overall. 
test           result  vs the others in its run  difficulty      counts  why
GSM8K (maker)   74.4%  █···█·██··█● beats 5/5    easy · avg 82%    6.9%  sharp · at its level
MATH (maker)    28.4%  █·█·█·██···● beats 5/5    hard · avg 32%    4.1%  typical
other tests (10) 
test                                result  vs the others in its run    difficulty           counts  why
ARC Challenge (maker)                59.7%  ▄···▄···█▄·● beats 5/5      easy · avg 77%         8.5%  sharp · at its level
MMLU (maker)                         70.6%  ▄····▄▄·▄··● beats 5/5      easy · avg 82%         8.1%  at its level
WinoGrande (maker)                   77.2%  █··█·█·█●··█ beats 4/5      easy · avg 86%         4.5%  at its level
Natural Questions (closed book) (…   30.6%  █·····▄▄▄··● beats 5/5      medium · avg 41%       4.4%  at its level
TriviaQA (maker)                     71.5%  █···██··█·●█ beats 4/5      easy · avg 83%         3.8%  at its level
HellaSwag (maker)                    84.4%  █···██···█●█ beats 4/5      very easy · avg 91%    3.0%  typical
Open PT-LLM Leaderboard average (…   67.9%  ▃▁▁▁▁▁▁▁▂●▅▁ beats 141/276  medium · avg 72%       1.8%  at its level · noisy
ARC Easy (maker)                     83.1%  ▄···▄···█▄·● beats 5/5      easy · avg 86%         1.6%  blunt
EuroEval Life in the UK (English …   87.2%  ▁▁▁▁▁▁▁▂▄●▄▃ beats 105/146  easy · avg 82%         1.2%  at its level · noisy
PIQA (maker)                         83.6%  ▄····▄··█▄·● beats 5/5      easy · avg 87%         1.1%  blunt · at its level
## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its *counts* column above; here it is added up by who ran the tests. [Method](../method.md) 
source                                    results  share of the score
Mixtral of Experts (submitted / maker-r…       12   62% ▇▇▇▇▇▇▇
Granite Code Models: A Family of Open F…        3   29% ▇▇▇▇
CRUXEval published evaluation results           2    5% ▇
Open Portuguese LLM Leaderboard                 1    2% ▇
EuroEval leaderboards                           1    1% ▇
prior: pull toward the middle                        0% ▇
weighted average                                   = 42.7
Licences: Mixtral of Experts ([CC BY 4.0](../../licenses/CC-BY-4.0.txt)) · Granite Code Models: A Family of Open Foundation Models for Code Intelligence ([CC BY 4.0](../../licenses/paper-2405.04324v1-CC-BY-4.0.txt)) · CRUXEval published evaluation results ([MIT](../../licenses/cruxeval-published-results-MIT.txt)) · Open Portuguese LLM Leaderboard ([Apache-2.0](../../licenses/hf-leaderboard-spaces-Apache-2.0.txt)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE)) ## How to run it ### llama.cpp (recommended) 1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md) ```sh # macOS / Linux curl -LsSf https://llama.app/install.sh | sh # Windows irm https://llama.app/install.ps1 | iex ``` 2. Run Mixtral 8x7B v0.1 (base) (Q4_K_M, needs ~26.4 GB, from [huggingface.co/TheBloke](https://huggingface.co/TheBloke/Mixtral-8x7B-v0.1-GGUF/tree/38762deaceb8f976428ab44e4c3ccf927a91132a)) ```sh llama cli -hf TheBloke/Mixtral-8x7B-v0.1-GGUF:Q4_K_M ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf TheBloke/Mixtral-8x7B-v0.1-GGUF:Q4_K_M ``` Other versions: Q8_0 (~49.6 GB) `llama cli -hf TheBloke/Mixtral-8x7B-v0.1-GGUF:Q8_0` ### LM Studio 1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download) 2. Run Mixtral 8x7B v0.1 (base) (Q4_K_M, needs ~26.4 GB, from [huggingface.co/TheBloke](https://huggingface.co/TheBloke/Mixtral-8x7B-v0.1-GGUF/tree/38762deaceb8f976428ab44e4c3ccf927a91132a)) ```sh lms get https://huggingface.co/TheBloke/Mixtral-8x7B-v0.1-GGUF@Q4_K_M -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: Q8_0 (~49.6 GB) `lms get https://huggingface.co/TheBloke/Mixtral-8x7B-v0.1-GGUF@Q8_0 -y` · 4bit (~26.5 GB, Mac only) `lms get https://huggingface.co/mlx-community/Mixtral-8x7B-v0.1-hf-4bit-mlx -y` ### Ollama 1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download) ```sh # Linux curl -fsSL https://ollama.com/install.sh | sh ``` 2. Run Mixtral 8x7B v0.1 (base) (Q4_K_M, needs ~28 GB, from [ollama.com](https://ollama.com/library/mixtral/tags)) ```sh ollama run mixtral:8x7b-text-v0.1-q4_K_M ``` Other versions: Q8_0 (~50 GB) `ollama run mixtral:8x7b-text-v0.1-q8_0` ### MLX (Mac) 1. Install MLX (Mac). Apple silicon Macs only, text only. ```sh # with uv uv tool install mlx-lm # or pip install mlx-lm ``` 2. Run Mixtral 8x7B v0.1 (base) (4bit, needs ~26.5 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/Mixtral-8x7B-v0.1-hf-4bit-mlx/tree/fc07711a141e4636987f2fba4ebd8ee59dc2a4fb)) ```sh mlx_lm.chat --model mlx-community/Mixtral-8x7B-v0.1-hf-4bit-mlx ``` 3. Or start an OpenAI-compatible server at http://localhost:8080/v1 ```sh mlx_lm.server --model mlx-community/Mixtral-8x7B-v0.1-hf-4bit-mlx ``` Model repository and documentation: [huggingface.co/mistralai](https://huggingface.co/mistralai/Mixtral-8x7B-v0.1) [interactive version](https://whatmodel.sh/model/mixtral-8x7b-v0-1) · [JSON](https://whatmodel.sh/api?format=json&all=1)