LLM Perks

Llama 3.2 90B Vision

Free right now Frontier
VisionTool useOpen weights

Llama 3.2 90B Vision is a language model with 90B parameters from Meta. It is currently free at 2 providers with up to 16K tokens of context. The model understands images, supports tool calling. Its weights are open, so it can also run locally.

Developer
Meta
Parameters
90B
Context
16K tokens
Type
Text
License
Open weights

Where to use Llama 3.2 90B Vision for free

Each row is a separate free endpoint with its own limits.

ProviderContextLimitCard
Together AI
nim/meta/llama-3.2-90b-vision-instruct
16K Fair use (depends on server load) — Open
NVIDIA Build
meta/llama-3.2-90b-vision-instruct
— Free serverless inference for development; limits and model availability may change. No Open

Quick start

curl https://api.together.xyz/v1/chat/completions \
  -H "Authorization: Bearer $TOGETHER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nim/meta/llama-3.2-90b-vision-instruct",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Example for Together AI. Any OpenAI-compatible client works: Cursor, Cline, OpenCode, Open WebUI.

Benchmarks

Intelligence Index
6.4
GPQA Diamond
43.2%
Humanity's Last Exam
4.5%

Source: Artificial Analysis · Llama 3.2 Instruct 90B (Vision)

Similar free models

FAQ

Is Llama 3.2 90B Vision free?

Yes. As of October 29, 2026, Llama 3.2 90B Vision is free at Together AI, NVIDIA Build. Each provider has its own limits — see the table above.

What is the context window of Llama 3.2 90B Vision?

The largest context among free routes is 16,384 tokens. Some providers cap it lower.

How do I call Llama 3.2 90B Vision via API?

Use any OpenAI-compatible client with base URL https://api.together.xyz/v1, model ID nim/meta/llama-3.2-90b-vision-instruct, and a Together AI key.

How good is Llama 3.2 90B Vision?

Artificial Analysis Intelligence Index: 6.4, GPQA Diamond: 43%. For comparison, the strongest free model in the catalog right now is GLM 5.3.

Is Llama 3.2 90B Vision open-weight?

Yes, the weights are published — you can download and run it locally, for example with Ollama or LM Studio.