Pixtral 12B + Pixtral Large: Mistral Multimodal — Resolution-Free + Apache 2.0

Pixtral 12B (Mistral Nemo 12B + 400M ViT) + Pixtral Large (124B) — Mistral'in açık multimodal'ı. Apache 2.0, resolution-free, EU AI Act-compliance friendly. 7-32 image per context, 128K context. RTX 4090'da Pixtral 12B QLoRA marjinal (~22 GB).

Şükrü Yusuf KAYA

22 dakikalık okuma

14.05.2026

İleri

Pixtral 12B + Pixtral Large: Mistral Multimodal — Resolution-Free + Apache 2.0

1. Pixtral Spec#

Model	Total	Vision	LLM Base	Context	Lisans
Pixtral 12B	12B + 400M ViT	Pixtral ViT 400M	Mistral Nemo 12B	128K	Apache 2.0
Pixtral Large	124B	Pixtral ViT 1B	Mistral Large 2 123B	128K	Mistral Research

Apache 2.0 önemli: EU AI Act + commercial production için en esnek lisans.

# Pixtral 12B basic FT
from transformers import LlavaForConditionalGeneration, AutoProcessor
model = LlavaForConditionalGeneration.from_pretrained(
    "mistral-community/pixtral-12b",
    quantization_config=bnb_4bit,
    torch_dtype="bfloat16",
)
# LoRA + Visual instruction tune workflow Qwen 2.5-VL ile aynı

✅ Teslim