The AIA Platform provides support for various AI models from different providers, ensuring flexibility for users to select the most suitable solution for their needs. The pricing structure depends on the type of model chosen and the provider’s specific terms. The platform supports two primary model types:
- Self-hosted models: Pricing is based on server costs and the computational resources required.
- Cloud-based models from third-party providers (e.g., OpenAI): Pricing is determined by token consumption rates defined by the provider.
The AIA Platform itself does not charge any additional fees for model usage. All costs are imposed directly by the model provider based on their pricing and API key terms.
- Token Analytics: Tracks the number of characters sent and processed by models in real-time.
- Flexible Integration: Supports multiple AI models, providing users with diverse options.
- Transparent Billing: Billing is directly tied to the provider's API key and pricing rules.
- Scenario-Based Pricing: Costs vary depending on the number of assistant users, the size of responses, and the complexity of the tasks.
For example, voice assistants (EAA) may incur variable token consumption based on the webpage size or the length of interactions.
¶ 3. Types of Models and Their Pricing
Self-Hosted Models:
- These models require users to maintain their own infrastructure.
- Pricing depends on server specifications:
- Computational power (e.g., CPU, GPU).
- Storage and memory usage.
- Operational uptime.
- Example cost: Hosting a model on a GPU-enabled server (e.g., NVIDIA A100) may cost $1–$3 per hour.
Cloud-Based Models (e.g., OpenAI):
- Pricing is based on the number of tokens processed:
- Input tokens: Data sent to the model.
- Output tokens: Data generated by the model in response.
- Example pricing for GPT-4o-2024-08-06:
- Input Tokens: $0.0015 per 1,000 tokens.
- Output Tokens: $0.0020 per 1,000 tokens.
- Token usage varies depending on the query size, user interaction complexity, and assistant configuration.
Token Distribution:
- Input Tokens: 11,773
- Output Tokens: 11,773
Cost Calculation:
- Input Tokens: Cost=11,773×0.00151,000=0.0177 USD\text{Cost} = \frac{11,773 \times 0.0015}{1,000} = 0.0177 \, \text{USD}Cost=1,00011,773×0.0015=0.0177USD
- Output Tokens: Cost=11,773×0.00201,000=0.0235 USD\text{Cost} = \frac{11,773 \times 0.0020}{1,000} = 0.0235 \, \text{USD}Cost=1,00011,773×0.0020=0.0235USD
- Total Cost: 0.0177+0.0235=0.0412 USD (4.12 cents)0.0177 + 0.0235 = 0.0412 \, \text{USD (4.12 cents)}0.0177+0.0235=0.0412USD (4.12 cents)
Stages of Interaction:
- Homepage: 30,000 tokens
- Category Page: 30,000 tokens
- Product Page: 30,000 tokens
- Cart: 30,000 tokens
- Checkout: 30,000 tokens
Total: 150,000 tokens
Token Distribution:
- Input Tokens: 75,000
- Output Tokens: 75,000
Cost Calculation:
- Input Tokens: Cost=75,000×0.00151,000=0.1125 USD\text{Cost} = \frac{75,000 \times 0.0015}{1,000} = 0.1125 \, \text{USD}Cost=1,00075,000×0.0015=0.1125USD
- Output Tokens: Cost=75,000×0.00201,000=0.15 USD\text{Cost} = \frac{75,000 \times 0.0020}{1,000} = 0.15 \, \text{USD}Cost=1,00075,000×0.0020=0.15USD
- Total Cost: 0.1125+0.15=0.2625 USD (26.25 cents)0.1125 + 0.15 = 0.2625 \, \text{USD (26.25 cents)}0.1125+0.15=0.2625USD (26.25 cents)
Token Distribution:
- Input Tokens: 64,000
- Output Tokens: 64,000
Cost Calculation:
- Input Tokens: Cost=64,000×0.00151,000=0.096 USD\text{Cost} = \frac{64,000 \times 0.0015}{1,000} = 0.096 \, \text{USD}Cost=1,00064,000×0.0015=0.096USD
- Output Tokens: Cost=64,000×0.00201,000=0.128 USD\text{Cost} = \frac{64,000 \times 0.0020}{1,000} = 0.128 \, \text{USD}Cost=1,00064,000×0.0020=0.128USD
- Total Cost: 0.096+0.128=0.224 USD (22.4 cents)0.096 + 0.128 = 0.224 \, \text{USD (22.4 cents)}0.096+0.128=0.224USD (22.4 cents)
- API Key Terms: By providing an API key, users accept the model provider’s pricing and usage terms.
- No AIA Platform Fees: The AIA Platform does not charge users directly for using models. Costs are solely dependent on the provider’s pricing.
AIA Platform includes a suite of tools to help users manage and predict costs:
- Token Analytics: Real-time tracking of input and output tokens.
- Usage Dashboard: Transparent billing insights based on model-specific pricing.
- Scenario Analysis: Allows users to simulate token usage for various scenarios (e.g., single queries, multi-step interactions).
The AIA Platform provides a clear and predictable pricing structure while allowing users to leverage a wide range of AI models. By offering robust analytics and direct billing tied to model providers, it ensures transparency and cost efficiency.
Whether you’re setting up a simple chatbot or an advanced voice assistant for EAA, the platform provides the tools to manage costs effectively while focusing on delivering exceptional user experiences.