readygpu

Gemma 4 12B

12B parameters · 256k context · apache-2.0 · released 2026-05-23

chatcodingvision

What to run it on in India

Fully in memory at 8k context, priced at retailers today.

Something we price holds it fully in memory, but none reaches 15 tok/s, our bar for recommending it. The table below has the speeds.

All hardware that runs it

Everything we price, best value first. Two cards split the model by layers.

HardwareQuantFitsDecodePricetok/s per 1,000
NVIDIA GeForce RTX 3090Q8_0fits44 tok/s––
NVIDIA GeForce RTX 5090Q8_0fits85 tok/s––
Mac Studio M4 Max 128GBQ8_0fits27 tok/s––
NVIDIA GeForce RTX 4090Q8_0fits48 tok/s––
NVIDIA GeForce RTX 5070 TiQ8_0fits42 tok/s––
Mac mini M4 Pro 64GBQ8_0fits14 tok/s––
NVIDIA GeForce RTX 5060 Ti 16GBQ8_0fits21 tok/s––
AMD Radeon RX 7900 XTXQ8_0fits40 tok/s––
Framework Desktop Ryzen AI Max+ 395 128GBQ8_0fits11 tok/s––
Mac Studio M3 Ultra 512GBQ8_0fits41 tok/s––
NVIDIA DGX Spark 128GBQ8_0fits13 tok/s––
NVIDIA GeForce RTX 3060 12GBQ6_Kfits22 tok/s––
AMD Radeon RX 9070 XTQ8_0fits27 tok/s––
GMKtec EVO-X2 Ryzen AI Max+ 395 128GBQ8_0fits11 tok/s––
Mac Studio M3 Ultra 256GBQ8_0fits41 tok/s––
Mac mini M4 Pro 48GBQ8_0fits14 tok/s––
NVIDIA GeForce RTX 4060 Ti 16GBQ8_0fits14 tok/s––
NVIDIA GeForce RTX 5080Q8_0fits45 tok/s––
Mac Studio M4 Max 64GBQ8_0fits27 tok/s––
NVIDIA GeForce RTX 4070 Ti SuperQ8_0fits32 tok/s––
NVIDIA GeForce RTX 5070Q6_Kfits41 tok/s––
AMD Radeon RX 9060 XT 16GBQ8_0fits13 tok/s––
Mac Studio M3 Ultra 96GBQ8_0fits41 tok/s––
NVIDIA GeForce RTX 4070 SuperQ6_Kfits31 tok/s––
NVIDIA GeForce RTX 4080 SuperQ8_0fits35 tok/s––
NVIDIA RTX PRO 6000 BlackwellQ8_0fits85 tok/s––
AMD Radeon AI PRO R9700Q8_0fits27 tok/s––
AMD Radeon RX 7800 XTQ8_0fits26 tok/s––
AMD Radeon RX 7900 XTQ8_0fits34 tok/s––
ASUS Ascent GX10 128GBQ8_0fits13 tok/s––
Framework Desktop Ryzen AI Max+ 395 64GBQ8_0fits11 tok/s––
Intel Arc B580Q6_Kfits22 tok/s––
Mac mini M4 32GBQ8_0fits6 tok/s––
NVIDIA GeForce RTX 3090 TiQ8_0fits48 tok/s––
AMD Radeon RX 7600 XTQ8_0fits12 tok/s––
Mac mini M4 16GBQ6_Kfits8 tok/s––
NVIDIA GeForce RTX 3080 10GBQ5_K_Mfits53 tok/s––
Intel Arc Pro B60Q8_0fits17 tok/s––
NVIDIA Tesla P40Q8_0fits16 tok/s––
Intel Arc A770 16GBQ8_0fits21 tok/s––

Memory needed

Weights, KV cache and runtime buffers at each context length (KV cache at fp16).

QuantFile@8k@32k@64k@128k
Q4_K_Mgguf6.6 GB7.8 GB8.3 GB9.0 GB10.4 GB
mlx-4bitmlx6.3 GB7.5 GB8.0 GB8.7 GB10.0 GB
Q5_K_Mgguf7.8 GB9.0 GB9.5 GB10.2 GB11.6 GB
Q6_Kgguf9.1 GB10.3 GB10.8 GB11.5 GB12.9 GB
Q8_0gguf11.8 GB13.0 GB13.5 GB14.2 GB15.6 GB
mlx-8bitmlx11.8 GB13.0 GB13.5 GB14.2 GB15.6 GB

Run it

Copy a command for the runtime you use.

  • Q4_K_M
    • Ollamaollama run hf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_M
    • LM Studiolms get unsloth/gemma-4-12b-it-GGUF@q4_k_m
    • llama.cppllama-server -hf unsloth/gemma-4-12b-it-GGUF:Q4_K_M
  • mlx-4bit
    • LM Studiolms get mlx-community/gemma-4-12B-it-4bit
    • MLXmlx_lm.server --model mlx-community/gemma-4-12B-it-4bit
  • Q5_K_M
    • Ollamaollama run hf.co/unsloth/gemma-4-12b-it-GGUF:Q5_K_M
    • LM Studiolms get unsloth/gemma-4-12b-it-GGUF@q5_k_m
    • llama.cppllama-server -hf unsloth/gemma-4-12b-it-GGUF:Q5_K_M
  • Q6_K
    • Ollamaollama run hf.co/unsloth/gemma-4-12b-it-GGUF:Q6_K
    • LM Studiolms get unsloth/gemma-4-12b-it-GGUF@q6_k
    • llama.cppllama-server -hf unsloth/gemma-4-12b-it-GGUF:Q6_K
  • Q8_0
    • Ollamaollama run hf.co/unsloth/gemma-4-12b-it-GGUF:Q8_0
    • LM Studiolms get unsloth/gemma-4-12b-it-GGUF@q8_0
    • llama.cppllama-server -hf unsloth/gemma-4-12b-it-GGUF:Q8_0
  • mlx-8bit
    • LM Studiolms get mlx-community/gemma-4-12B-it-8bit
    • MLXmlx_lm.server --model mlx-community/gemma-4-12B-it-8bit