Talk to AI with audio, image or video Runs on our GPU
Show it, play it, ask it. Send a recording, a picture or a short video and get a written or spoken answer.
5 credits · about 5 min on our GPU with Image in · spoken answerQuickest: Short video in · text answer, 3 credits
Qwen2.5-Omni understands speech, images and short videos in one model. Ask a question about what you send and get the answer as text, or as text plus a spoken reply.
What you can do
Each example is a real run on our GPU: what went in, the exact instruction, and the result exactly as the model made it. Pick one to see it, then start from the same instruction.
What we asked
Look at this chart and tell me, as if speaking to a colleague, what changed around week 23 and what the peak downlink value is.
What came out, untouched
Well, around week 23, there was a big change. You see, the 5G - SA was switched on. And the peak downlink value? It's about 195 GB. If you have any more questions about this chart or anything else, feel free to ask.
4 min 11 s on our RTX 4060 5 credits ($0.50) · Image in · spoken answer
A short video clip (play it under "Prompt and settings we used" in the gallery).
What we asked
Describe this video in three sentences: the scene, the motion, and the mood.
What came out, untouched
The scene is a snowy mountain landscape at night, with a campfire glowing warmly in the foreground. The motion is subtle, with a person walking towards the fire, their silhouette visible against the bright flames. The mood is serene and peaceful, with the vast starry sky above and the snow-covered mountains creating a sense of tranquility. What do you think about this kind of scene?
1 min 24 s on our RTX 4060 3 credits ($0.30) · Short video in · text answer
2 examples from our own runs, with the times we measured. Open "Prompt and settings we used" under any of them for the exact input.
Chart question, answered aloud
Week-23 switch-on and ~195 GB peak, spoken in the model's built-in voice.
The picture that went in
Prompt
Look at this chart and tell me, as if speaking to a colleague, what changed around week 23 and what the peak downlink value is.
The scene is a snowy mountain landscape at night, with a campfire glowing warmly in the foreground. The motion is subtle, with a person walking towards the fire, their silhouette visible against the bright flames. The mood is serene and peaceful, with the vast starry sky above and the snow-covered mountains creating a sense of tranquility. What do you think about this kind of scene?
Describe a short video
Scene and mood right, but it invented 'a person walking towards the fire': check video answers.
The video that went in
Prompt
Describe this video in three sentences: the scene, the motion, and the mood.
The prompt and the model decide the output; bigger models and settings cost more credits. These notes come from what worked, and what did not, in our runs.
Times are end to end on our GPU (an RTX 4060 Laptop GPU, 8 GB), model loading included, before any jobs ahead of yours. "Estimate" marks an option we have not measured yet.
| Option | Credits | Typical time | Slowest seen | What to expect |
|---|---|---|---|---|
| Image in · spoken answerRecommended | 5 | 4 min 51 s | 4 min 51 s1 run | Measured once: a chart question answered correctly, aloud, in 251 s plus 40 s load. |
| Short video in · text answer | 3 | 2 min 4 s | 2 min 4 s1 run | Measured once: 84 s plus 40 s load. Scene and mood right, but it invented a person walking. |
| Audio in · transcript and spoken answer | 17 | 16 min 48 s | 16 min 48 s1 run | Measured once: transcript word for word, but slow: 972 s plus 36 s load (the model overflows the 8 GB card). |
You answer for your prompt and your choice of model and settings; the safety filter still checks every job, and a refused prompt costs nothing.
5 creditsabout 5 min on our GPU, plus any jobs ahead
You'll have 7 days to download it; then it's deleted.
Results are kept 7 days. Credits come back only if the job fails or never starts.
Reading the line…
Credits are held when you send a job and spent when the result is ready. They come back automatically if the job fails, is refused or has not started within 24 hours. Results are kept 7 days, or 48 hours after your first download, then deleted; a result that expires unused is not refunded. Pricing and refunds
One image, one voice recording or one short video per job, with a question.
Our measured audio job took about 17 minutes of GPU time against 5 for an image.
Yes, the image and audio options return a spoken reply as well as text.
Transcripts and chart answers were right in our trials; one video description invented a person. Check before relying on it.
Ask about an image is faster and cheaper for pictures; use this one for audio, video or a spoken answer.