---

medgemma-27b-text-it-GGUF

Original Model

google/medgemma-27b-text-it

Run with LlamaEdge

• LlamaEdge version: v0.18.5 and above

• Prompt template

  • Prompt type: gemma-3

  • Prompt string

    <bos><start_of_turn>user
    {user_message}<end_of_turn>
    <start_of_turn>model
    {model_message}<end_of_turn>model
    

• Context size: 128000

• Run as LlamaEdge service

  • Chat

    wasmedge --dir .:. --nn-preload default:GGML:AUTO:medgemma-27b-text-it-Q5_K_M.gguf \
      llama-api-server.wasm \
      --prompt-template gemma-3 \
      --ctx-size 128000 \
      --model-name medgemma-27b
    

  • Images

    Note that input images are required to be normalized to 896 x 896 resolution and encoded to 256 tokens each

    wasmedge --dir .:. --nn-preload default:GGML:AUTO:medgemma-27b-text-it-Q5_K_M.gguf \
      llama-api-server.wasm \
      --prompt-template gemma-3 \
      --llava-mmproj medgemma-27b-text-it-mmproj.gguf \
      --ctx-size 128000 \
      --model-name medgemma-27b
    

• Run as LlamaEdge command app

  wasmedge --dir .:. \
    --nn-preload default:GGML:AUTO:medgemma-27b-text-it-Q5_K_M.gguf \
    llama-chat.wasm \
    --prompt-template gemma-3 \
    --ctx-size 128000
  

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
medgemma-27b-text-it-Q2_K.gguf	Q2_K	2	10.5 GB	smallest, significant quality loss - not recommended for most purposes
medgemma-27b-text-it-Q3_K_L.gguf	Q3_K_L	3	14.5 GB	small, substantial quality loss
medgemma-27b-text-it-Q3_K_M.gguf	Q3_K_M	3	13.4 GB	very small, high quality loss
medgemma-27b-text-it-Q3_K_S.gguf	Q3_K_S	3	12.2 GB	very small, high quality loss
medgemma-27b-text-it-Q4_0.gguf	Q4_0	4	15.6 GB	legacy; small, very high quality loss - prefer using Q3_K_M
medgemma-27b-text-it-Q4_K_M.gguf	Q4_K_M	4	16.5 GB	medium, balanced quality - recommended
medgemma-27b-text-it-Q4_K_S.gguf	Q4_K_S	4	15.7 GB	small, greater quality loss
medgemma-27b-text-it-Q5_0.gguf	Q5_0	5	18.8 GB	legacy; medium, balanced quality - prefer using Q4_K_M
medgemma-27b-text-it-Q5_K_M.gguf	Q5_K_M	5	19.3 GB	large, very low quality loss - recommended
medgemma-27b-text-it-Q5_K_S.gguf	Q5_K_S	5	18.8 GB	large, low quality loss - recommended
medgemma-27b-text-it-Q6_K.gguf	Q6_K	6	22.2 GB	very large, extremely low quality loss
medgemma-27b-text-it-Q8_0.gguf	Q8_0	8	28.7 GB	very large, extremely low quality loss - not recommended
medgemma-27b-text-it-f16-00001-of-00002.gguf	f16	16	30.0 GB
medgemma-27b-text-it-f16-00002-of-00002.gguf	f16	16	24.1 GB
medgemma-27b-text-it-mmproj.gguf	f16	16	858 MB

Quantized with llama.cpp b5452