Model
Model explorer

Llama 3.2 11B Vision

OPEN
Meta · Llama 3.2 family · released Sep 25, 2024

First Llama vision model, adding image reasoning/captioning on an 8B-text backbone (English-only for image+text).

ReasoningCodingVisionFunction callingTool useAgentic
682.7
Elo · rank #268
Parameters
10.6B
Active params
10.6B (dense)
Context
128K tokens
Architecture
Dense decoder-only Transformer + cross-attention vision adapter (GQA)
License
Llama 3.2 Community License Agreement
Languages
1+
API price (in/out)
No hosted API
Modalities
text · vision
Benchmark results
Bar shows position within the tracked field; marker = field best
AI2DVision91.1%#20
best: Molmo 72B · 96.3%
ChartQAVision83.4%#25
best: MiniMax-VL-01 · 91.7%
DocVQAVision88.4%#37
best: Qwen2-VL-72B · 96.5%
GPQA DiamondReasoning32.8%#250
best: GPT-5.6 · 94.6%
MATH-500Math51.9%#124
best: GPT-5 · 99.4%
MathVistaVision51.5%#51
best: Seed 2.1 Pro · 90.7%
MGSMMath68.9%#36
best: OpenAI o4-mini · 93.7%
MMLUKnowledge73.0%#131
best: OpenAI o3 · 92.9%
MMMU-ProVision23.7%#51
best: Claude Opus 4.7 · 85.5%
MMMUVision50.7%#93
best: Claude Fable 5 · 89.3%
SEED-BenchVision72.6%#3
best: NAVER HyperCLOVA X SEED Think 32B · 77.9%
VQAv2Vision75.2%#15
best: Reka Edge (2026) · 88.4%
Run it locally
VRAM @ Q4
7.5 GB
VRAM @ FP16
21 GB
Fits on (Q4)
RTX 3060 12GBRTX 4070 Ti 16GBRTX 3090 24GBRTX 4090 24GBRTX 5090 32GBM4 Pro 48GBM3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB
llama.cpp's multimodal (mmproj/mtmd) vision support only matured in 2025; no widely reported RTX 4090 Q4 tok/s figure was found
Quantizations
GGUF Q4 · AWQ · GPTQ · MLX
Llama 3.2 family
Elo progression across releases
Llama 3.2 11B VisionSep 2024682.7Llama 3.2 1BSep 202481.3Llama 3.2 3BSep 2024380.2Llama 3.2 90B VisionSep 2024999.5
Resources
HFmeta-llama/Llama-3.2-11B-Vision-InstructGHmeta-llama/llama-modelsDOCSllama.com/docs/model-cards-and-prompt-formats/llama3_2APIAvailable via Meta API
API price weights · each benchmark row carries its own source badge (see methodology)