OpenAI API

About OpenAI API

OpenAI API Platform is the developer product family for building software on OpenAI models and tools. It provides programmatic access to frontier models, multimodal input and output, tool use, realtime voice, image generation, embeddings and other developer capabilities through supported APIs and SDKs. This page treats those capabilities as parts of one developer platform instead of separate software products unless OpenAI markets a durable named product line with its own buyer intent.

What is included in the OpenAI API Platform?

The platform centers on the Responses API and supported SDKs, with access to current frontier and specialized models. Developers can build text and reasoning workflows, multimodal applications, image generation and editing, speech and transcription experiences, realtime voice systems, retrieval and embedding workflows, and tool-using agents.

OpenAI changes individual model names and snapshots frequently. Buyers should therefore evaluate the platform around durable capabilities, supported APIs, model lifecycle policy, latency, governance and cost rather than treating every model snapshot or documentation category as a separate product.

How is API pricing structured?

Pricing checked on September 8, 2026 against OpenAI's current API Platform pages. OpenAI bills model usage primarily by tokens or other model-specific units. GPT-6 Astra is currently listed at $10 per 1 million input tokens and $50 per 1 million output tokens. GPT-5.6 Sol is listed at $4 input and $20 output per 1 million tokens, while Terra and Luna provide lower-cost options.

Tool calls, image generation, realtime audio and other specialized capabilities can have separate charges. Teams should model total workload cost using expected input size, output size, caching, tool calls and traffic volume instead of comparing only one headline token price.

Where do GPT-6 Astra and Realtime API fit?

GPT-6 Astra is a current model product line available through the API and is kept as a separate Brandligo product because OpenAI released it under a durable name with its own model page, pricing and migration guidance. Realtime API is also kept separately because it is a named developer product for low-latency voice and realtime interaction.

By contrast, embeddings, image generation, text and reasoning categories, and older Whisper-oriented speech pages are treated here as capabilities or model categories inside the API Platform rather than independent software products.

How should teams choose models and endpoints?

Teams should start with the workload rather than the model name. Complex reasoning, computer use and long-running end-to-end tasks may justify a frontier model such as GPT-6 Astra. High-volume production workloads can favor lower-cost models. Voice agents need realtime latency and audio support, while retrieval systems may depend on embeddings and file search.

A production evaluation should measure quality on representative tasks, latency, retry behavior, tool reliability, context size, cost and failure modes. Model selection should be revisited when OpenAI changes the lineup or pricing.

What are the main governance and deployment considerations?

Organizations should review data handling, retention, authentication, rate limits, regional requirements, model access controls and application-level permissions before deployment. Tool-using agents can act on external systems, so least-privilege credentials and approval boundaries matter as much as prompt quality.

Teams should also plan for model migrations. OpenAI publishes model deprecations and migration guidance, and production systems should avoid assuming a particular snapshot will remain available indefinitely.

Who should choose something else?

A different platform may be better when a company needs a model provider that is already standardized inside its cloud contract, requires a specific open-weight deployment model, or needs infrastructure controls that are not available in the OpenAI API. Some workloads also benefit from specialist speech, search, vision or coding services.

The OpenAI API Platform is strongest for teams that value access to OpenAI's current models and developer tooling and are prepared to manage usage, model changes and application governance as part of the software lifecycle.