Atlassian–OpenAI Rovo Deal: Pricing Impact

Atlassian and OpenAI expanded their partnership on October 6, 2026. OpenAI says GPT-6-family models will power agents across Atlassian’s platform and Rovo, the assistant that connects company work through its Teamwork Graph. The deal also gives Atlassian broader access to GPT-6 Astra and the GPT-5.6 series. It does not announce a new public API tariff or the price Atlassian will pay OpenAI.

What changed—and what is still exploratory

Rovo can draw on Jira tickets, Confluence documents, discussions, and other work context to answer questions such as whether a launch is on track. OpenAI says Atlassian can bring newer reasoning models into Rovo through its APIs as the models advance. The companies also point to Atlassian and Teamwork Graph plugins that bring relevant project information into ChatGPT and Codex, subject to permissions.

OpenAI says more than 3,000 Atlassian developers use Codex across terminals, IDEs, and code review. It also says the companies are exploring deeper Jira integrations to assign work to AI agents, track progress, capture decisions, and review results. Exploring is not a release date or a promise that every customer can use those flows today.

Announcement detail	Status	Budget implication
OpenAI models in Rovo and Atlassian agents	Expanded agreement	Evaluate Rovo usage under Atlassian terms
GPT-6 Astra and GPT-5.6 access	Named in OpenAI’s announcement	No customer-facing discount disclosed
ChatGPT/Codex plugins for Atlassian context	Existing integration described	Check ChatGPT/Codex seats and connector permissions
Deeper Jira agent delegation	Exploration	Do not budget as a shipped feature

Pricing impact: no new public rate announced

The live table, chart, and calculator above show direct OpenAI API list rates for a separately built agent. They are not Rovo prices or Atlassian’s confidential contract. OpenAI’s announcement gives no contract value, token volume, negotiated discount, Rovo-credit change, or new ChatGPT Enterprise price.

Atlassian’s Rovo licensing page says Rovo credits are charged by the work performed, not whether the action starts in Jira, Confluence, Slack, or another supported surface. Paid Atlassian plans contribute a shared monthly Rovo-credit allowance based on product, edition, and seat count. That is a different meter from an application calling OpenAI’s API directly. A ChatGPT Enterprise or Codex arrangement is another contract surface again.

For the current model list, see our OpenAI pricing page. Compare a second model supplier on our Anthropic pricing page, and run your own input/output mix through the token calculator. Our OpenAI API pricing guide explains cache and long-context costs that a simple per-token headline misses.

Who benefits, who may pay more

Atlassian-heavy teams may benefit from AI working with work-item history, documentation, and permissions already in their tools. The business case is less time reconstructing context and more accepted actions—not a promised lower token bill.

Buyers with large Rovo or agent fleets risk treating this as “free GPT-6 inside Jira.” Neither company says that. Credit allowances, overages, seat commitments, permission reviews, and human approval time can dominate the practical cost. Frontier-model routing may also cost more than a smaller model on routine summaries; the announcement does not disclose Atlassian’s routing logic.

Developers can use the partnership as a prompt to test connector quality: does the agent retrieve the right Jira issue and Confluence decision, respect access controls, cite its evidence, and make only authorized changes? A model’s raw token price does not measure those failures.

For teams comparing a separately metered coding-agent option, review the Z.ai coding plan as a control. It is not a Rovo substitute; compare it only on the coding tasks it can actually perform.

Affiliate disclosure: AI Pricing Guru may earn a commission from the sponsored link at no extra cost to you.

What to do before expanding use

1. Separate Rovo usage, ChatGPT/Codex seats, and direct API calls in the budget. Ask each vendor which meter applies to the exact workflow.
2. Run a pilot on representative Jira-to-Confluence and coding tasks. Capture credit use or billed tokens, cache hits, retries, latency, reviewer minutes, and accepted outcomes.
3. Test permissions and write approvals with real roles. Measure the cost of wrong-context answers and unauthorized-action prevention, not just successful demos.
4. Request the current Rovo allowance, overage rules, model availability, data-handling terms, and any enterprise quote in writing before forecasting a rollout.

Bottom line: the partnership is a deeper distribution and workflow deal for OpenAI models inside Atlassian. The price question remains product-specific; today’s announcement is not a public price cut.

Sources: OpenAI’s October 6 partnership announcement, API pricing documentation, and Atlassian’s Rovo licensing page. Checked October 6, 2026.