Description
️ 🖼Tool Name:
AgentKit
🔖 Categories:
Automation and Smart Agents
Integrations and APIs
Programming and Development
No-Code Workflows
Testing and Quality Assurance
️ ✏What does this tool offer?
AgentKit is a comprehensive suite of tools from OpenAI designed for developers and organizations to build, deploy, and optimize intelligent agents.
The suite enables visual workflow design through Agent Builder,management of data sources and tools via the Connector Registry, and integration of agent-based conversational experiences into applications and websites using ChatKit.
It also includes capabilities for evaluating agent performance, such as datasets, execution path evaluation, and automated prompt optimization, in addition to support for third-party AI models.
AgentKit also provides Reinforcement Fine-Tuning capabilities for customizing inference models, along with features such as custom tool calls and custom evaluators.
According to the information provided, ChatKit and the new Evals capabilities were available to all developers, while Agent Builder was available in beta, and the Connector Registry began rolling out in beta to select API, ChatGPT Enterprise, and Edu customers.
In an update dated June 3, 2026, OpenAI announced that it would phase out the Agent Builderand Evalsproducts, ceasing to make them available on the OpenAI platform as of November 30, 2026. OpenAI recommended using the Agents SDK for “code-as-workflow” scenarios andWorkspace Agents in ChatGPT for use cases that rely on natural language instructions.
⭐ What does it actually offer based on user experience?
- Building intelligent agents and developing their workflows.
- Visually design workflows using Agent Builder.
- Managing data sources and tools using the Connector Registry.
- Integrating agent-based conversational experiences into apps and websites via ChatKit.
- Evaluate agent performance using datasets and evaluate execution paths.
- Automatically optimize claims.
- Use third-party AI models.
- Customize inference models using Reinforcement Fine-Tuning.
- Create custom tool and evaluator callbacks.
🤖 Does it include automation?
Yes. AgentKit focuses on building intelligent agents and their workflows, providing tools to visually design these workflows, manage tools and data sources, and evaluate and optimize agent performance. The Agents SDK can also be used for workflows that run as code.
💰 Pricing Model:
It is based on the usage cost of the OpenAI API;AgentKit does not have a separate monthly subscription. Some components of AgentKit were previously available at no additional cost under the OpenAI API usage pricing model.
🆓 Free Plan Details:
| Service | Details |
|---|---|
| AgentKit | No standalone free plan listed |
| ChatKit | Was available to all developers |
| Evals | Was available to all developers |
| Agent Builder | Was available as a beta |
| Connector Registry | Began a pilot rollout to select API, ChatGPT Enterprise, and Edu customers |
💳 Paid plan details:
| Plan / Service | Price | Features / Details |
|---|---|---|
| AgentKit | No separate monthly subscription listed | The pricing model is based on the cost of using OpenAI API models |
| OpenAI API usage | Depends on usage | Cost is calculated based on usage of OpenAI API models |
| Some AgentKit components | are included at no additional charge | Some components were included in the OpenAI API usage pricing |
🧭 How to access the tool:
| How to access | Details |
|---|---|
| API | Available through the OpenAI API |
| Software Development | The Agents SDK can be used for workflows that run as code |
| Web / Applications | Agent-based conversational experiences can be integrated into apps and websites using ChatKit |
🔗 Link to the demo or official website:
OpenAI — Introducing AgentKit
