How to Add AI Video Generation to Your Mobile App: A Developer's Guide

How to Add AI Video Generation to Your Mobile App: A Developer's Guide

Imagine: a user clicks 'Generate Video' and gets a ready clip in the app within a minute. But in reality, models like Runway Gen-3, Sora, Kling require powerful GPUs (A100/H100) and take 30 seconds to several minutes. The mobile developer's task is to set up an async flow so the user doesn't leave while waiting. We integrate these APIs into your iOS/Android app, ensuring reliable generation, caching, and notifications.

How to Integrate Runway API into a Mobile App?

Runway provides a REST API with polling for status checks. On iOS, we implemented a service with async/await and a progress bar that simulates execution based on typical generation time. On Android, we use WorkManager with CoroutineWorker—the task runs in the background even when the app is minimized. Here's an example iOS service:

class VideoGenerationService {
    func generate(prompt: String, sourceImage: UIImage?) async throws -> URL {
        let taskId = try await runwayClient.createTask(
            prompt: prompt,
            imageURL: sourceImage.map { try await uploadImage($0) },
            duration: 5,
            ratio: "1280:768"
        )
        UserDefaults.standard.set(taskId, forKey: "pendingVideoTaskId")
        return try await pollWithBackoff(taskId: taskId)
    }

    private func pollWithBackoff(taskId: String) async throws -> URL {
        let intervals: [TimeInterval] = [3, 5, 8, 10, 10, 15, 15, 20, 20, 30]
        for interval in intervals + Array(repeating: 30.0, count: 10) {
            try await Task.sleep(nanoseconds: UInt64(interval * 1e9))
            let task = try await runwayClient.getTask(id: taskId)
            switch task.status {
            case .succeeded:
                UserDefaults.standard.removeObject(forKey: "pendingVideoTaskId")
                return task.output.first!
            case .failed:
                throw VideoGenError.generationFailed(task.failure ?? "Unknown")
            default: continue
            }
        }
        throw VideoGenError.timeout
    }
}

On Android: WorkManager with CoroutineWorker is the right choice for long-running background tasks. Polling in doWork(), Result.retry() on PROCESSING, Result.success(outputData) on SUCCEEDED.

What to Do If Generation Takes Longer Than a Minute?

Push notification is a must-have option. The backend tracks task status and sends FCM/APNs on completion. Deep links (Universal Links / App Links) lead to the result screen with auto-play. If the user minimized the app, progress is saved via pendingTaskId in UserDefaults.

Estimating Real Progress Without API Data

Most APIs do not return a percentage—only status PENDING/PROCESSING/SUCCEEDED. We use a simulated progress bar: a timer for 55 seconds (95%), then wait for the actual response. This beats an empty spinner.

AI Video API Comparison Table

Runway API is about 2x faster than Kling AI for typical clips. To choose an API, consider generation time and clip length. Runway API is the most mature with SDKs for TypeScript/Python. Sora from OpenAI is currently only available through a partner program. Compliance with the App Store Review Guidelines (section 4.2) is mandatory for publication.

Comparison of Async Generation Approaches

The choice between polling and WebSocket depends on update frequency and criticality of progress. For simple scenarios, polling suffices; for complex ones, WebSocket.

Deliverables & What's Included

• Analysis: selecting the optimal API for your needs (clip length, budget, quality).
• Integration: implementing the async flow with polling, progress bar, and result caching.
• Background processing: WorkManager/BGTaskScheduler for long tasks.
• Push notifications: FCM/APNs with deep linking.
• Documentation: usage guide and API documentation.
• Access: repository access and credentials for all services.
• Training: up to 2 hours of knowledge transfer for your team.
• Support: 3 months of free post-launch support.

Our Work Process

1. Analysis — we study your app, select an API, estimate load.
2. Design — architecture of the flow, caching scheme, security.
3. Implementation — writing code, integrating SDK, testing.
4. Testing — verification on real devices, error simulation.
5. Deployment — publication to App Store / Google Play, monitoring setup.

Timeline and Cost

Basic integration (one API, player, cache) takes 5 to 7 days and costs around $3,000 to $5,000. Full flow with background tasks, push, Image-to-Video, and gallery takes 3 to 4 weeks and costs $10,000 to $15,000. The exact cost is calculated individually after reviewing the project.

Our experience integrating AI features spans more than 5 years and 15+ successful projects. We guarantee quality and compliance with the App Store Review Guidelines (section 4.2).

Ready to start? Contact us for a preliminary estimate. Get a consultation on API selection and architecture today.

Machine Learning in Mobile Apps: CoreML, TFLite, and On-Device Models

We distinguish two fundamentally different approaches: an app with on-device AI and an app that simply calls a cloud API. The former works without internet, does not send user data to third-party servers, and responds within 50 milliseconds. The latter depends on network latency and pricing plans. Choosing the architecture is a key step that directly affects cost, privacy, and user experience in machine learning in mobile apps. Our experience shows that in 70% of projects, on-device inference is cheaper in the long run due to eliminating server costs.

How to Choose Between CoreML and TFLite for On-Device Inference?

CoreML — Apple's native framework for running ML models on device. Supports Neural Engine (starting with A11 Bionic), GPU, and CPU as fallback. Models are converted to .mlmodel format via coremltools from PyTorch, ONNX, or TensorFlow. Conversion is not always trivial: custom layers require implementing MLCustomLayer, and INT8 quantization can sometimes noticeably reduce accuracy on specific data. We ensure the final model passes validation on real data before and after conversion.

TensorFlow Lite — cross-platform alternative for Android and Flutter. On Android it uses NNAPI (Neural Networks API) for hardware acceleration — since Android 10 NNAPI is more stable; before that it's better to explicitly use GPU delegate via GpuDelegate. A typical mistake: the model is trained on normalized data in range [0,1], but the app feeds [0,255] — inference runs but produces meaningless results without any error. We include an automatic input data validation module in the SDK.

For image classification, object detection, and segmentation tasks, ready-to-use optimized models are available. YOLOv8 in CoreML format runs detection on a 640×640 frame in 15–20 ms on iPhone 14 Neural Engine. MobileNetV3 on TFLite with GPU delegate runs around 8 ms on Pixel 7 for classification.

What If You Need Text Generation On-Device?

Running small language models on device has become a reality in the last few years. Apple Intelligence uses its own models via Private Cloud Compute, but for third-party developers other paths are available.

llama.cpp with Metal backend on iOS is a working approach for phi-3-mini (3.8B parameters, 4-bit quantization, ~2.3 GB). Inference: 15–25 tokens/second on iPhone 15 Pro. For integration in Swift, use the Swift Package llama.swift or a wrapper via C interface llama.h. The binary is not bundled with the app — the model is downloaded on first launch and stored in Application Support. Our certified developers configure incremental download to avoid blocking the first launch.

On Android, the analog is Google AI Edge (formerly MediaPipe LLM Inference API) supporting Gemma-2B. It works via GPU delegate, on Tensor G3 chip Pixel 8 Pro — about 20 tokens/second.

Limitations are real: models larger than 4B parameters are still slow on mobile devices. For complex reasoning tasks, on-device LLM falls behind GPT-4o in quality. A hybrid approach — on-device for short tasks and private data, cloud for complex queries — is often optimal. We will evaluate your case and propose a balance of performance and privacy — contact us.

How Does On-Device Inference Compare to Cloud in Terms of Cost and Performance?

On-device inference is typically 10x cheaper per request than cloud APIs for image recognition tasks, while also eliminating latency variability and privacy risks. The table below summarizes the trade-offs.

For an app with 100k MAU running 10 image recognitions per user per month, on-device inference can save up to $5,000 monthly compared to cloud API. Get a free consultation on your ML architecture today.

Integrating OpenAI API and Other Cloud Models

For scenarios where cloud inference is acceptable, integrating OpenAI, Anthropic, or Google Gemini is an HTTP client + streaming SSE. In Swift, AsyncThrowingStream is convenient for streaming responses. In Kotlin, use Flow.

Critically: API keys must never be stored in the app bundle. Even an obfuscated key can be extracted from the IPA in 10 minutes using strings or frida. Correct architecture: mobile app → your own backend → OpenAI API. The backend controls rate limiting, logs requests, and protects the key.

What Is Included in the Work (Deliverables)

• Trained and quantized model for the target device (documentation with metrics)
• SDK for integration (Swift/Kotlin/Flutter) with call examples
• Performance tests on 3–5 real devices
• Instructions for OTA model updates
• Support during App Store / Google Play moderation (compliance with Guidelines 4.2, 5.1)
• 2 weeks of technical support after release

Typical Project Pipeline

1. Task analysis — measure latency, privacy, size, supported devices.
2. Model prototyping — in Python, evaluate accuracy on target data.
3. Conversion and quantization — for CoreML/TFLite with validation.
4. Integration into the app — model wrapped in a service layer (easy to swap CoreML ↔ TFLite ↔ cloud).
5. Testing — on real devices, measure FPS, RAM, battery.
6. Deployment — via TestFlight / Firebase App Distribution, monitor metrics.

Timelines: integration of a ready CoreML/TFLite model — 1–2 weeks, development of a custom model with mobile optimization — from 6 weeks, on-device LLM chat with personalization — 4–8 weeks.

Why We Take on Complex Cases?

10+ years of experience in mobile development, 50+ implemented AI/ML solutions, guarantee of compatibility with current iOS and Android versions. All projects undergo code review and load testing. The cost includes preparation of moderation documentation and training of your team.

Contact us — we will help you choose the architecture and implement ML in your app turnkey. Order an audit of your existing solution — we will assess the potential for server cost savings free of charge. In some projects, savings can reach significant amounts per month.