Measured on 3 providers. The model card lists text, image, video input; what each provider accepts is below.
How much of your video will Nemotron 3.5 Super VL see? Try the calculator →
Each limit as a percentage of the best provider's for this model, averaged over the limits that provider was measured on. Video density counts as video tokens per minute; tokens per image are left out. 100 means it matches the best provider on every limit. Providers measured on fewer than two limits aren't scored. How it's scored.
| Model | Provider | Value |
|---|---|---|
| Nemotron 3.5 Super VL | Together AI | 100 |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | 100 |
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | 75 |
Every route got the same 5-minute 1080p video. Video tokens per minute is what the provider billed for it: the amount of visual input the model got, and what ranks providers. Frames per minute is how many frames that was, and how large, worked out from the bills and the model's own preprocessing code.
"On the first minute", "on a 360p copy": the provider couldn't take the original file whole, so it got the first part or a smaller copy of the same video. That counts.
Fixed per video: the provider sends the same number of frames however long the video is, so longer videos get fewer per minute.
Bars are scaled within each model. Hover a row for how the frames were worked out. How we measure.
| Model | Provider | Video tokens / min | Frames / min | Frame size | Measured on |
|---|---|---|---|---|---|
| Nemotron 3.5 Super VL | Together AI | 9,470 | 120 | 672×384 | the first minute |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | 9,470 | 120 | 672×384 | the first minute |
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | 900 | 6.2 (fixed per video) | the 5-minute clip |
768×432 frames, the size most video pipelines send. Whichever limit hits first counts: image count, decoded pixels or the context window.
| Model | Provider | Value |
|---|---|---|
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | 32 |
| Nemotron 3.5 Super VL | Together AI | 32 |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | 32 |
Tiny 64×64 images, so only the count limit is tested. Real frames can hit other limits sooner; see Frames per request.
| Model | Provider | Value |
|---|---|---|
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | 32 |
| Nemotron 3.5 Super VL | Together AI | 32 |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | 32 |
A video sent as a video_url. Most of these models only take images; those show No video.
| Model | Provider | Value |
|---|---|---|
| Nemotron 3.5 Super VL | Together AI | 256 MB |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | 256 MB |
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | ≥ 96 MB? |
Base64 images inside the request. Capped at 16 MB by our probe; ≥ means it accepted everything we sent.
| Model | Provider | Value |
|---|---|---|
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | ≥ 32 MB |
| Nemotron 3.5 Super VL | Together AI | ≥ 32 MB |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | ≥ 32 MB |
Images sent as links the provider fetches itself. Capped at 128 MB by our probe.
| Model | Provider | Value |
|---|---|---|
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | ≥ 128 MB |
| Nemotron 3.5 Super VL | Together AI | ≥ 128 MB |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | ≥ 128 MB |
Prompt tokens per 768×432 frame. Lower is cheaper: a 100-frame clip costs 100× this.
| Model | Provider | Value |
|---|---|---|
| Nemotron 3.5 Super VL | Together AI | 345 |
| Nemotron 3.5 Super VL | Together AI via OpenRouter | 345 |
| Nemotron 3.5 Super VL | DeepInfra (fp4) via OpenRouter | 340 |
Measured 2026-11-08
Measured 2026-11-08
Measured 2026-11-08