OpenAI Free Workspace Agents Extended to July 6, 2026
Quick Answer: OpenAI has extended the free period for workspace agents from May 6 to July 6, 2026. After that date, credit-based pricing takes effect for ChatGPT Business, Enterprise, and Edu users. A typical agent run using GPT-5.5 will cost between 5 and 25 credits.
If your team has been relying on OpenAI workspace agents for automated workflows across Slack, Salesforce, or Google Drive, you just got a reprieve. OpenAI quietly extended the free trial period from the original May 6 deadline to July 6, 2026. That is two extra months of production-level AI agent usage at no charge, and it gives teams valuable time to understand exactly what credit-based pricing will look like once the billing switch flips on July 6.
What exactly are OpenAI workspace agents and why do they matter?
OpenAI launched workspace agents on April 22, 2026 as the replacement for Custom GPTs within ChatGPT Business, Enterprise, and Edu plans. Unlike standalone chatbots, workspace agents are deeply integrated into your organization’s ecosystem. They can pull data from Salesforce, post updates to Slack, analyze Google Drive documents, and execute multi-step workflows that previously required separate automation tools. The original announcement positioned them as free until May 6 before transitioning to credit-based pricing. The release notes for ChatGPT Business and Enterprise from early June 2026 confirmed the extension to July 6, 2026, giving adopters significantly more runway.
Workspace agents represent a strategic shift for OpenAI. Instead of selling the feature as a flat-rate add-on, the company is moving toward usage-based billing. That matters because it changes how teams budget for AI tools. A team that runs fifty agent workflows per day will pay differently than one running five. The extension gives organizations the chance to measure their actual usage patterns before the meter starts running.
What does the new credit-based pricing look like?
OpenAI has published a rate card for ChatGPT Business, Enterprise, and Edu customers that outlines how workspace agent credits will work after July 6. According to the card, a typical end-to-end workspace agent run using GPT-5.5 may consume between 5 and 25 credits. The exact credit cost depends on the model used, the complexity of the task, the number of tool calls made, and whether the run uses fast mode or standard processing.
Credits are purchased as a pool that your organization draws from. This is similar to how Codex moved to token-based credit billing in April 2026, where lighter tasks now cost less than they did under flat per-message pricing. For teams that have been building heavily on workspace agents during the free period, now is the time to audit your actual usage and project what your monthly credit consumption will look like after July 6.
How does this compare to other AI pricing changes in 2026?
OpenAI is not alone in rethinking how it charges for AI agents. 2026 has been a year of significant pricing shifts across the industry. Anthropic announced credit-based billing for Claude agents on June 15, ending agent subsidies. Google dropped its AI Ultra tier from $250 to $200 at I/O 2026 and introduced Gemini Spark, a free tier for developers. xAI rolled out Grok Build 01 with agentic coding API access at competitive rates.
The common thread is that AI providers are moving away from unlimited flat-rate agent access toward usage-based models. The era of “all you can eat” AI agents is ending. OpenAI’s workspace agent extension gives teams two more months of free access, but the direction is clear. For a comprehensive breakdown of how the major providers compare, check the AI subscription tiers comparison.
How should teams prepare for the July 6 transition?
The two-month extension is a gift of time, but it should not be wasted. Here are the steps every team using workspace agents should take before July 6:
Audit your current usage — Track how many workspace agent runs your team performs per day, what models they use, and how many tool calls each run makes. This data will let you estimate your monthly credit needs.
Prioritize high-value workflows — Identify which agent automations deliver the most ROI. Low-impact experiments may not be worth the credits after billing starts.
Explore free alternatives — Google’s Gemini API free tier offers 60 requests per minute on certain models. Open-source agent frameworks like Microsoft’s recently released open-source agent framework or OpenClaw provide self-hosted options with no per-query costs.
Set credit budgets — Establish per-user or per-team credit limits to avoid surprise bills. OpenAI’s rate card supports admin controls for credit allocation.
Test during the free period — Use the remaining weeks to test edge cases, load test your automations, and gather baseline metrics while it costs nothing.
What happens to workspace agents running in Slack and other integrations?
Workspace agents invoked outside of ChatGPT (for example, within Slack or Salesforce) will have different credit consumption rates than those invoked inside ChatGPT. OpenAI’s rate card documentation explicitly notes this distinction. If your team runs agents primarily through Slack or other embedded integrations, those runs may cost more credits than the same task initiated from within ChatGPT.
This is an important detail for teams that adopted workspace agents specifically for Slack-based workflows. The ScriptWalker analysis of the workspace agents launch noted that OpenAI intentionally priced external integrations higher to encourage users to interact through the ChatGPT interface. Teams should account for this differential when estimating post-July costs.
Is there a Codex-only seat option for teams that only need agentic coding?
Yes. As of April 2, 2026, OpenAI introduced flexible pricing with a new Codex-only seat type for ChatGPT Business and Enterprise plans. This seat costs less than a full ChatGPT Business seat and provides access specifically to Codex for agentic coding without the full ChatGPT suite. This is a strategic option for development teams that primarily need coding agents rather than general workspace agents. The Codex-only seat does not unlock workspace agents, but it does give teams access to OpenAI’s agentic coding environment at a reduced per-user cost.
According to OpenAI’s help center, on average Codex costs approximately $100-$200 per developer per month, though actual costs vary significantly based on model selection, number of instances, automations, and use of fast mode. Teams should compare the Codex-only seat pricing against their projected workspace agent credit costs to find the most cost-effective combination.
🔑 Key Takeaways
OpenAI extended the free period for workspace agents from May 6 to July 6, 2026, giving Business and Enterprise teams two extra months of no-cost access to agent workflows across Slack, Salesforce, and Google Drive.
Credit-based pricing begins on July 6, with a typical GPT-5.5 agent run costing between 5 and 25 credits depending on complexity, model, tool calls, and whether it is invoked inside or outside ChatGPT.
Teams should audit their current agent usage patterns during the free extension to build accurate budget projections for the credit-based billing model that takes effect July 6.
External integrations like Slack-based agents will cost more credits than the same task performed within ChatGPT, according to OpenAI’s published rate card.
Free alternatives exist including Google’s Gemini API free tier, open-source agent frameworks from Microsoft and others, and self-hosted tools that may reduce or eliminate per-query costs for teams with simpler needs.
Frequently Asked Questions
When does the free period for OpenAI workspace agents end?
The free period now ends on July 6, 2026, extended from the original May 6, 2026 deadline. Credit-based pricing will begin on that date for ChatGPT Business, Enterprise, and Edu customers. The extension was confirmed in the June 2026 release notes for both Business and Enterprise editions.
What are OpenAI workspace agents?
Workspace agents are AI assistants within ChatGPT Business and Enterprise that can perform tasks across integrated apps like Slack, Salesforce, and Google Drive. They launched on April 22, 2026 as the replacement for Custom GPTs and allow teams to build automated, AI-powered workflows without separate automation tools.
How much will OpenAI workspace agents cost after the free period?
Credit-based pricing takes effect July 6. A typical end-to-end workspace agent run using GPT-5.5 consumes between 5 and 25 credits. The exact cost depends on the model, number of tool calls, use of fast mode, and whether the agent is invoked inside ChatGPT or from an external integration like Slack.
Can free-tier ChatGPT users access workspace agents?
No, workspace agents are exclusive to ChatGPT Business, Enterprise, and Edu plans. Free, Plus, Go, and Pro users do not have access to workspace agents. The feature is designed for organizational use with shared workspaces, admin controls, and team-level permission management.
What alternatives exist if I don’t want to pay for workspace agents?
Teams looking for free AI workflow alternatives can consider Google’s Gemini API with its generous free tier offering 60 requests per minute, open-source agent frameworks like Microsoft’s recently released framework, or self-hosted options like LocalAI and OpenClaw. Each trades ease of setup for reduced or eliminated per-query costs.