---
title: "Llama 3.2 (Vision) 90B Benchmarks - RankedAGI"
description: "Detailed benchmark and metadata record for Llama 3.2 (Vision)."
source: "https://rankedagi.com/models/llama-3.2-(vision)-90b"
---

# Llama 3.2 (Vision)

| Field | Value |
| --- | --- |
| Organization | Meta |
| License | Open Source |
| Size | 90B |
| Version | Latest |
| Released | 2024-09-25 |
| Context window | 128K |
| Knowledge cutoff | 2023-12-01 |
| Input cost per million tokens | $0.9 |
| Output cost per million tokens | $0.9 |
| Last updated | 2026-04-11T19:46:04.657Z |

## Benchmarks

| Benchmark | Category | Value | Description | Source |
| --- | --- | --- | --- | --- |
| RankedAGI Coding | coding | 40.0% | RankedAGI Coding Score |  |
| RankedAGI Agentic | Agents | 45.2% | RankedAGI Agentic Score |  |
| RankedAGI Reasoning | reasoning | 41.4% | RankedAGI Reasoning Score |  |
| GPQA Diamond | reasoning | 43.0% | Generalized Prefix Question Answering Score (Reasoning) PhD Level Reasoning |  |
| MMLU | knowledge | 86.0% | Image Reasoning |  |
| Hughes Hallucination Evaluation | safety | 4.3% | Computed using Vectara's Hallucination Evaluation Model, also known as HHEM. This evaluates how often an LLM introduces hallucinations when summarizing a document. | https://github.com/vectara/hallucination-leaderboard |
| Artificial Analysis Quality Index | general | 67 | Artificial Analysis Quality Index (third-party composite, deprecated) |  |
| RankedAGI Math | math | 46.4% | RankedAGI Math Score |  |
| RankedAGI Overall | general | 44.6% | Overall RankedAGI score |  |