vlm.fyi by Ambient Get the data
Providers

Together AI via OpenRouter

4 models measured.

How much of a video does the model actually see?

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.

Video tokens / min billed for the video; ranks providers Frames delivered / min and the size each frame was resized to
Muse Spark (open) 15,601 on the first minute 120 at 588×336 · on the first minute Nemotron 3.5 Super VL 9,470 on the first minute 120 at 672×384 · on the first minute

+ 2 providers did not accept video: Kimi K3 and Kimi K3.1 Pro+ 2 providers did not accept video: Kimi K3 and Kimi K3.1 Pro+ 2 providers did not accept video: Kimi K3 and Kimi K3.1 Pro

"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.

Bars are scaled within each model. Hover a row for how the frames were worked out. How we measure.

Show as table
ModelProviderVideo tokens / minFrames / minFrame sizeMeasured on
Muse Spark (open)Together AI via OpenRouter 15,601120 588×336the first minute
Nemotron 3.5 Super VLTogether AI via OpenRouter 9,470120 672×384the first minute
Kimi K3Together AI via OpenRouter No video
Kimi K3.1 ProTogether AI via OpenRouter No video

How Together AI via OpenRouter compares with other providers

Each dot is one or more of the model's other providers on that limit; the highlighted dot is Together AI via OpenRouter. The overall score ranks a model's providers: each limit as a percentage of the best provider's, averaged over the limits measured. 100 means it matches the best provider on every limit.

Kimi K3against its 11 other providers

Overall score?623rd of 12 scored · 6 limitsbest 67 · Modal
Images724best ≥ 1,024 · Modal
768×432 frames≥ 384best, tied
Inline body≥ 32 MBbest, tied
Images by URL≥ 128 MBbest, tied
Video fileNo videobest ≥ 512 MB · Decart (mxfp4) via OpenRouter
Video tok/minNobest 7,886 · Moonshot AI

Kimi K3.1 Proagainst its 10 other providers

Overall score?549th of 11 scored · 6 limitsbest 67 · Fireworks
Images724best 999 · Together AI
768×432 frames≥ 384best, tied
Inline body≥ 32 MBbest, tied
Images by URL64 MBbest ≥ 128 MB · DeepInfra (bf16) via OpenRouter
Video fileNo videobest ≥ 512 MB · Together AI
Video tok/minNobest 324,240 · Fireworks

Muse Spark (open)against its 5 other providers

Overall score?952nd of 6 scored · 6 limitsbest 100 · Together AI
Images724best ≥ 1,024 · Together AI
768×432 frames128best, tied
Inline body≥ 32 MBbest, tied
Images by URL≥ 128 MBbest, tied
Video file256 MBbest, tied
Video tok/min15,601best, tied

Nemotron 3.5 Super VLagainst its 2 other providers

Overall score?1001st of 3 scored · 6 limitsbest, tied
Images32best, tied
768×432 frames32best, tied
Inline body≥ 32 MBbest, tied
Images by URL≥ 128 MBbest, tied
Video file256 MBbest, tied
Video tok/min9,470best, tied

Kimi K3

moonshotai/kimi-k3
Images
724 “This endpoint's maximum context length is 1048576 tokens. However, you requested about 1049347 tokens (1706 of text input, 1047625 of image input, 16 in the output). Please reduce the length of either one, or use the context-compression plugin to compress your prompt automatically.”
768×432 frames
≥ 384
Tokens / frame
457
Inline body
≥ 32 MB
Images by URL
≥ 128 MB
Video file
No video “OpenRouter had no endpoint to route this to (removed by its Input Video Support filter).”
Video tok/min
No “OpenRouter had no endpoint to route this to (removed by its Input Video Support filter).”
Frames/min
—
Video input
No
Tool messages
Strict order
Extra fields
All accepted

Measured 2026-11-04

Kimi K3.1 Pro

moonshotai/kimi-k3.1-pro
Images
724 “This endpoint's maximum context length is 1048576 tokens. However, you requested about 1049347 tokens (1706 of text input, 1047625 of image input, 16 in the output). Please reduce the length of either one, or use the context-compression plugin to compress your prompt automatically.”
768×432 frames
≥ 384
Tokens / frame
457
Inline body
≥ 32 MB
Images by URL
64 MB “Failed to start generation: Failed to generate completions: CannotConnect (a request carrying 128 MB of images by URL)”
Video file
No video “OpenRouter had no endpoint to route this to (removed by its Input Video Support filter).”
Video tok/min
No “OpenRouter had no endpoint to route this to (removed by its Input Video Support filter).”
Frames/min
—
Video input
No
Tool messages
Strict order
Extra fields
All accepted

Measured 2026-11-08

Muse Spark (open)

meta/muse-spark-open
Images
724 “This endpoint's maximum context length is 1048576 tokens. However, you requested about 1049347 tokens (1706 of text input, 1047625 of image input, 16 in the output). Please reduce the length of either one, or use the context-compression plugin to compress your prompt automatically.”
768×432 frames
128 “Multimodal payload is 274341888 bytes (261.6 MiB) which exceeds the gRPC single-message ceiling of 267386880 bytes (255.0 MiB). Reduce the number of images, lower their resolution (e.g. `detail=”
Tokens / frame
408
Inline body
≥ 32 MB
Images by URL
≥ 128 MB
Video file
256 MB “Multimodal processing failed: Failed to finalize multimodal tracker: HTTP error while fetching media: error decoding response body”
Video tok/min
15,601
Frames/min
120
Video input
Yes
Tool messages
Strict order
Extra fields
All accepted

Measured 2026-11-08

Nemotron 3.5 Super VL

nvidia/nemotron-3.5-super-vl
Images
32 “Invalid multimodal request: invalid media request for model nemotron_vl: model spec nemotron_vl supports at most 32 image inputs; got 33”
768×432 frames
32 “Invalid multimodal request: invalid media request for model nemotron_vl: model spec nemotron_vl supports at most 32 image inputs; got 36”
Tokens / frame
345
Inline body
≥ 32 MB
Images by URL
≥ 128 MB
Video file
256 MB “Could not load the media in this request -- video #1: video input payload exceeds the maximum size of 268435456 bytes.”
Video tok/min
9,470
Frames/min
120
Video input
Yes
Tool messages
Strict order
Extra fields
All accepted

Measured 2026-11-08