OpenAI Model List: API Models and How to Choose

OpenAI API model list

This is a selection guide to OpenAI API models, not a list of models in the ChatGPT app. OpenAI’s API model catalog groups models by purpose. The table highlights flagship models listed in the catalog and specialized offerings described in OpenAI API announcements; it is not exhaustive. Catalog listing, an API announcement, and availability to a particular project are different kinds of evidence. Check the model list for your deployment project before using an ID in a request.

Model or API offering	Documented API model ID	When to consider it	Evidence and availability note
GPT-6 Astra	gpt-6-astra	OpenAI highlights computer use, browsing, software engineering, cybersecurity, science, and professional work. These are stated capabilities, not a guarantee of results on your workload.	The model catalog lists this ID. Its launch announcement described a limited rollout; the material cited here does not establish whether that restriction has since changed. Confirm availability for your project.
GPT-6.1 Sol	gpt-6.1-sol	A lower-priced candidate for complex work. OpenAI says it nearly matches Astra on agentic coding, computer use, and professional work at one-fifth of Astra’s standard input and output token prices.	The model catalog lists this ID. Project availability should still be checked.
GPT-6.1 Terra	gpt-6.1-terra	Consider for visual reasoning, computer use, and clear, natural writing.	The model catalog lists this ID. Project availability should still be checked.
GPT-6 Luna	gpt-6-luna	OpenAI positions Luna for focused, high-volume work where efficiency matters.	The model catalog lists this ID. Project availability should still be checked.
GPT-Image-2.6 Flare and Sunburst	Exact request IDs not established by the announcement cited here	Image generation and editing. OpenAI positions Flare for faster iteration and Sunburst for more detail and consistency at medium or high quality settings.	OpenAI announced these as API offerings. The cited announcement does not establish their exact request IDs or current project availability; confirm both in the model list and model documentation before integrating.
GPT-Live-1	Exact request ID not established by the announcement cited here	Real-time voice. OpenAI describes full-duplex conversation and interruption handling.	OpenAI announced this as an API offering. The cited announcement does not establish its exact request ID or current project availability; confirm both in the model list and model documentation before integrating.

The all-models catalog also groups specialized offerings into categories such as cybersecurity and life sciences, as well as image, realtime, transcription, and embedding models. This guide omits specific names in those specialty categories where the cited material does not establish an exact model ID and its access conditions. A product name in an announcement is not necessarily the exact string to send in an API request.

How to choose a model

Start with the task and constraints rather than assuming there is one best model:

• Complex reasoning or coding: Evaluate gpt-6-astra when capability is the priority. Compare it with gpt-6.1-sol when cost matters; OpenAI’s comparison supports that trade-off for specific evaluations, not a prediction of performance on every application.
• Higher-volume focused tasks: Consider gpt-6-luna as a candidate. Test it against your quality requirements and estimate cost using current rates and your expected input and output tokens.
• Visual reasoning or computer use: Evaluate gpt-6.1-terra for these documented use cases. Test it with the actual interface, inputs, and failure cases in your workflow.
• Live speech conversations: Consider GPT-Live-1 for a voice application. OpenAI describes full-duplex conversation and interruption handling; verify its exact model ID, project availability, and supported interface before integration.
• Image generation or editing: Compare Flare, positioned for faster iteration, with Sunburst, positioned for greater detail and consistency at higher quality settings. Verify the exact model IDs and current project availability before integration.

These are starting points based on OpenAI’s published positioning, not independent benchmarks. For production, run representative tasks through a shortlist and compare acceptance criteria, error cases, latency, and estimated cost. Include difficult inputs, not only examples where a model is likely to succeed.

Verify a model before production use

Use OpenAI’s documented model-list endpoint with credentials for the API project you plan to deploy. The API reference documents GET /v1/models and says the response lists currently available models with basic information, including model IDs and any announced shutdown date.

curl https://api.openai.com/v1/models \
  -H "Authorization: Bearer $OPENAI_API_KEY"

Check the returned data[].id values for the exact request identifier. If a shutdown_date is present, account for it in your plan. Project settings can control model access, so check with credentials associated with the intended project—not a different project or a ChatGPT model picker.

The list endpoint helps you check which model IDs are returned for those credentials; it does not describe every model’s capabilities or supported inputs and outputs. Consult the individual model documentation for those details. Before deployment, confirm documented features, any access conditions, and current lifecycle information, then test the model on representative requests.

Sources and update note

Catalog details and availability can change. Recheck the OpenAI API model catalog, all-models catalog, List models API reference, and documentation for the specific model before deployment. The model-list response is an availability check for the credentials used; it is not a capability guide or a guarantee that a model is suitable for your workload.

Sources

Models | OpenAI API

All models | OpenAI API

List models | OpenAI API Reference

Managing projects in the API platform

GPT-6 Astra: A new generation of intelligence

Introducing GPT-6.1 Sol

Introducing GPT-6.1 Terra

Build more natural voice experiences with GPT-Live-1 in the API

Introducing Images 2.6