metadata

license: apache-2.0
model_name: Mistral-Small-3.1-24B-Instruct-2503
base_model: mistralai/Mistral-Small-3.1-24B-Instruct-2503
model_creator: mistralai
quantized_by: Second State Inc.
inference: false
language:
  - en
  - fr
  - de
  - es
  - pt
  - it
  - ja
  - ko
  - ru
  - zh
  - ar
  - fa
  - id
  - ms
  - ne
  - pl
  - ro
  - sr
  - sv
  - tr
  - uk
  - vi
  - hi
  - bn
tags:
  - transformers
pipeline_tag: image-text-to-text

Mistral-Small-3.1-24B-Instruct-2503-GGUF

Original Model

mistralai/Mistral-Small-3.1-24B-Instruct-2503

Run with LlamaEdge

LlamaEdge version: v0.16.5 and above

Prompt template

Chat

Prompt type: mistral-small-chat

Prompt string

<s>[SYSTEM_PROMPT]<system prompt>[/SYSTEM_PROMPT][INST]<user message>[/INST]<assistant response></s>[INST]<user message>[/INST]

Context size: 32000

Run as LlamaEdge service

Chat

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Mistral-Small-3.1-24B-Instruct-2503-Q5_K_M.gguf \
    llama-api-server.wasm \
    --prompt-template mistral-small-chat \
    --ctx-size 32000 \
    --model-name Mistral-Small

Run as LlamaEdge command app

Chat

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Mistral-Small-3.1-24B-Instruct-2503-Q5_K_M.gguf \
  llama-chat.wasm \
  --prompt-template mistral-small-chat \
  --ctx-size 32000

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Mistral-Small-3.1-24B-Instruct-2503-Q2_K.gguf	Q2_K	2	8.89 GB	smallest, significant quality loss - not recommended for most purposes
Mistral-Small-3.1-24B-Instruct-2503-Q3_K_L.gguf	Q3_K_L	3	12.4 GB	small, substantial quality loss
Mistral-Small-3.1-24B-Instruct-2503-Q3_K_M.gguf	Q3_K_M	3	11.5 GB	very small, high quality loss
Mistral-Small-3.1-24B-Instruct-2503-Q3_K_S.gguf	Q3_K_S	3	10.4 GB	very small, high quality loss
Mistral-Small-3.1-24B-Instruct-2503-Q4_0.gguf	Q4_0	4	13.4 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Mistral-Small-3.1-24B-Instruct-2503-Q4_K_M.gguf	Q4_K_M	4	14.3 GB	medium, balanced quality - recommended
Mistral-Small-3.1-24B-Instruct-2503-Q4_K_S.gguf	Q4_K_S	4	13.5 GB	small, greater quality loss
Mistral-Small-3.1-24B-Instruct-2503-Q5_0.gguf	Q5_0	5	16.3 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Mistral-Small-3.1-24B-Instruct-2503-Q5_K_M.gguf	Q5_K_M	5	16.8 GB	large, very low quality loss - recommended
Mistral-Small-3.1-24B-Instruct-2503-Q5_K_S.gguf	Q5_K_S	5	16.3 GB	large, low quality loss - recommended
Mistral-Small-3.1-24B-Instruct-2503-Q6_K.gguf	Q6_K	6	19.3 GB	very large, extremely low quality loss
Mistral-Small-3.1-24B-Instruct-2503-Q8_0.gguf	Q8_0	8	25.1 GB	very large, extremely low quality loss - not recommended
Mistral-Small-3.1-24B-Instruct-2503-f16.gguf	f16	16	47.2 GB

Quantized with llama.cpp b4944.