vision-language
Molmo2-4B Video QA: Answer questions about video content with temporal understanding. Open-source vision-language model. Ready-to-use REST API, no cold starts, duration-based pricing.
Idle
Your request will cost $0.005 per run.
For $1 you can run this model approximately 200 times.
ExamplesView all
README
Molmo2 Video QA
Molmo2 Video QA is a powerful video understanding model that answers questions about video content. Simply upload a video and ask anything — the model analyzes visual scenes, actions, objects, and context to deliver accurate, natural-language responses.
Built for developers and creators who need intelligent video comprehension without building complex pipelines.
Why Choose This?
-
Natural language understanding Ask questions in plain English about what happens in your video — no need for timestamps or frame-by-frame annotation.
-
Scene and action recognition Understands objects, people, movements, environments, and temporal sequences across the video.
-
Flexible video input Accepts video uploads or public URLs for seamless integration into existing workflows.
-
Fast and accurate Optimized for quick turnaround while maintaining high comprehension accuracy.
-
Production-ready API Ready-to-use REST endpoint with predictable per-second pricing and no cold starts.
How to Use
- Upload your video — drag and drop a file or paste a public video URL.
- Write your question — describe what you want to know about the video content.
- Submit — the model processes the video and returns a natural-language answer.
- Iterate — ask follow-up questions or upload new videos as needed.
Pricing
Per-5-second billing with a 5-second minimum.
| Video Duration | Cost |
|---|---|
| Up to 5s | $0.005 |
| 10s | $0.01 |
| 30s | $0.03 |
| 60s | $0.06 |
| 120s (max) | $0.12 |
Billing Rules
- Minimum charge: 5 seconds ($0.005)
- Rate: $0.001 per second ($0.005 per 5 seconds)
- Maximum video length: 120 seconds (2 minutes)
Best Use Cases
- Content moderation — Automatically review video uploads for policy compliance.
- Video search and indexing — Extract semantic information for searchable video libraries.
- Accessibility — Generate descriptions of video content for visually impaired users.
- Education and training — Analyze instructional videos and answer learner questions.
- Surveillance and monitoring — Summarize events or detect specific actions in footage.
- Social media analytics — Understand trends and content themes across video posts.
Notes
- If using a URL, ensure it is publicly accessible. A preview thumbnail in the interface confirms successful access.
- For videos longer than 2 minutes, split into segments and process separately.
- Clear, well-lit footage with minimal background noise yields the best results.
- Be specific in your questions for more precise answers.