Available Models and Pricing
Overview
LearningFlow’s AI orchestration lets you choose from a wide variety of AI models and tools based on your specific needs—ranging from text generation, image creation, speech processing, to web crawling and more. Each of these AI capabilities is provided by different models or services, each with its own pricing structure.
Note: The latest, most up-to-date list of available AI models and their credit costs is maintained here.
You can always click to check the current offerings and rates.
Understanding the Model Pricing Table
Below is a simplified explanation to help you understand the key columns and terms you will see in the pricing and model list:
| Term | What It Means |
|---|---|
| Model Name | The name or identifier of the AI model or service (e.g., GPT-4 Turbo, Claude 2, Whisper-1). |
| Unit Type | The billing unit used—such as “tokens” for text, “calls” for image generations, “minutes” or “characters” for audio or speech services. |
| Currency Rate Input | The number of units included or processed per input request before charge calculation. |
| Currency Rate Output | The number of units included or processed per output generated. |
| Total Rate | The combined cost (in basic currency units) for processing the input plus output, representing the price for one full request cycle. |
| Currency Round Policy | Indicates if the usage is rounded or billed exactly. “False” means exact billing; “True” means rounding applies. |
| Details | Extra metadata or special flags about the model or usage. |
What You Need to Know
-
Different AI Models Serve Different Purposes:
Some models, like GPT-4 or Claude, generate human-like text; others generate images (DALL·E), transcribe speech (Whisper), or analyze images (Mathpix, Vision AI). -
Units & Billing:
Most text AIs measure usage in tokens (pieces of words). You pay per token processed, including both your input (prompt) and the AI's output (response).
For images, you pay per call to the model generating the image.
For speech, you pay by minutes of audio or number of characters processed. -
Cost Efficiency vs. Capabilities:
Models vary widely in power and cost. More advanced or larger models cost more but may give better results. Balancing quality and cost is key. -
Planning Your Usage:
Estimate token counts or usage ahead to manage costs. You can monitor credits and set budgets in your organization.
Model Examples and When to Use Them
| Model Name | Typical Usage | Approximate Cost per Use |
|---|---|---|
| GPT-3.5-turbo | General-purpose text generation & conversation | Moderate (~$0.0025 for 1K tokens) |
| GPT-4-turbo | More advanced, complex reasoning and understanding | Higher than GPT-3.5 |
| Claude-2 | Alternative LLM for conversational AI | Similar to GPT models |
| DALL·E-3 | Text-to-image generation | $0.03 per call |
| Whisper-1 | Speech-to-text transcription | $0.17 per audio minute |
| TTS-1 | Text-to-speech conversion | Charged per number of characters |
| Mathpix | Image-based text recognition (OCR) for math content | $0.05 per image |
How to Read Your Usage and Cost
- Each AI node in your flow consumes credits based on:
- The input size (e.g., how many tokens or length of audio).
- The output size (e.g., AI-generated tokens or audio length).
- For example, sending a 1000-token prompt to GPT-4 and receiving a 500-token output will cost credits for 1500 tokens total.
- Images always have fixed pricing per call.
- Speech services depend on processed minutes or characters.
Tips for Managing Your AI Usage
- Choose appropriate models: Use simpler models for non-critical tasks to save credits.
- Optimize prompts: Be concise but informative to reduce token usage.
- Pre-test flows to estimate credit consumption.
- Monitor usage regularly via the dashboard.
- Set budgets or limits in your organization to avoid surprise charges.
Detailed Model Overview
GPT-3.5-turbo and GPT-4 Turbo
These powerful language models support a wide range of text generation and comprehension tasks. Ideal for conversational bots, content creation, and educational question generation.
Anthropic Claude Models
Competitors to GPT, known for safety and ethical considerations, useful in chat and knowledge retrieval scenarios.
Vision and Image Models (DALL·E, GPT-4 Vision)
Generate images from text prompts or analyze images for content, widely used for creative assignments.
Speech Models (Whisper, TTS-1, ElevenLabs)
Convert speech-to-text and text-to-speech; great for language learning, accessibility, and interactive voice activities.
Where to Learn More
If you need help deciding which model fits your use case or want guidance on cost-effective design, reach out in our support forum.