Llama-3.2-90B-Vision-Instruct
The Llama 90B Vision model is a top-tier, 90-billion-parameter multimodal model designed for the most challenging visual reasoning and language tasks.
text → text · no single maker — the market carries it
Sold by 7 ways
| Seller | Lane | Rate |
|---|---|---|
| IBM watsonx cloud | standard | $2per Mtok in$2per Mtok outraw.githubusercontent.com · read 2026-09-16 |
| IBM watsonx.ai cloud | standard | $2per Mtok in$2per Mtok outraw.githubusercontent.com · read 2026-08-28 |
| Oracle Cloud Infrastructure cloud | standard | $2per Mtok in$2per Mtok outraw.githubusercontent.com · read 2026-09-16 |
| Microsoft Azure AI cloud | standard | $2.04per Mtok in$2.04per Mtok outraw.githubusercontent.com · read 2026-09-16 |
| DeepInfra aggregator | standard | $0.35per Mtok in$0.4per Mtok outapi.deepinfra.com · read 2026-08-25 |
| IO.NET aggregator | standard | $0.35per Mtok in$0.4per Mtok out$0.17per Mtok cached$0.7per Mtok cache writemodels.dev · read 2026-09-16 |
| Requesty aggregator | deepinfra | $0.35per Mtok in$0.4per Mtok out$0.35per Mtok cachedrouter.requesty.ai · read 2026-08-28 |
Measured 7 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 20thof 36 | Balrog — Epoch AI | Average progress | 0.273 |
| 26thof 217 | Mmlu — Epoch AI | EM | 0.803 |
| 46thof 46 | Enigma eval — Epoch AI | Accuracy | 0.38 |
| 74thof 108 | Math level 5 — Epoch AI | mean_score | 39.435 |
| 263rdof 313 | GPQA Diamond | mean_score | 41.035 |
| 268thof 290 | OTIS Mock AIME 2024-2025 | mean_score | 2.639 |
| 557thof 656 | Epoch capabilities index — Epoch AI | ECI Score | 125.5 |
About
Llama-3.2-90B-Vision-Instruct — a text model, sold by 7 companies from $0.35 in and $0.4 out per million tokens, placed 26th of 217 on Mmlu — Epoch AI.
It takes text and returns text, with a context window of 32,768 tokens. It was published in September 2024, trained on material up to December 2023. Its sellers say it can call a tool. The catalogue files it under reasoning. Seven companies sell it. The cheapest is $0.35 in and $0.4 out per million tokens at DeepInfra. Beside the standard rate there is a separately routed lane. It has been measured on 7 boards, and stands best at 26th of 217 on Mmlu — Epoch AI.
Every current figure
- Register
- model
- Takes
- text
- Returns
- text
- Context
- 32,768 tokens
- Longest answer
- 8,192 tokens
- Published
- September 2024
- Knowledge to
- December 2023
- Parameters
- 90 billion · read from its own name
- Licence
- not read
- Sellers
- 7
- Price
- $0.35 in and $0.4 out per million tokens — DeepInfra
- Boards
- 7
- Best place
- 26th of 217 — Mmlu — Epoch AI
Known as 2 names
meta-llama/Llama-3.2-90B-Vision-Instructdeepinfra/meta-llama/Llama-3.2-90B-Vision-Instruct