OpenAI Rolls Out GPT-6.1 Terra Across ChatGPT Work, Codex and the API
GPT-6.1 Terra expands across ChatGPT Work, Codex and the OpenAI API with lower-cost positioning for visual reasoning and computer-use tasks.
OpenAI has rolled out GPT-6.1 Terra, a model positioned around enhanced visual reasoning, precise computer-use capabilities and clearer writing. It is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, while developers can call it through the OpenAI API as gpt-6.1-terra. The release matters because OpenAI is pairing broad access with a substantially lower per-task cost claim, potentially making more AI-assisted operational work practical to test and scale.
According to OpenAI's official GPT-6.1 Terra announcement, the model is intended for work that combines visual information with actions and written output. OpenAI cites examples including interpreting technical drawings, drafting proposals and editing videos. Its API documentation also identifies catalog matching and form interpretation as relevant use cases.
The important distinction is not simply that another model is available. Terra is being presented as a lower-cost option for tasks where teams need visual understanding and computer use, rather than as a general replacement for every model or workflow. OpenAI says it outperforms GPT-5.6 Terra on key visual-reasoning tasks and approaches GPT-6 Astra on those tasks at a fraction of the cost. Those are OpenAI performance and cost-positioning claims, so businesses should validate them against their own inputs, tools and quality requirements.
Access, pricing and the practical opportunity
Developers can use the API model identifier gpt-6.1-terra. OpenAI lists standard API pricing of $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens. Cached input pricing is particularly relevant for repeated context, such as recurring reference material, catalog data or standard form layouts.
| Area | GPT-6.1 Terra | Relevant comparison in OpenAI's announcement |
|---|---|---|
| Access | ChatGPT Work, Codex and the OpenAI API as gpt-6.1-terra |
Broad multi-channel rollout is the announced availability change |
| Input pricing | $1 per million tokens, or $0.05 per million cached input tokens | OpenAI says per-task cost is substantially lower than GPT-5.6 Terra and GPT-6.1 Sol |
| Output pricing | $5 per million tokens | OpenAI positions Terra as approaching GPT-6 Astra on cited visual-reasoning tasks at a fraction of the cost |
| Primary focus | Visual reasoning, computer use and clearer writing | OpenAI says it outperforms GPT-5.6 Terra on key visual-reasoning tasks |
For businesses, lower token prices do not automatically equal lower operating costs. The useful measure is cost per completed, acceptable task. A workflow that requires extensive retries, manual correction or complex surrounding software can cost more than a higher-priced model that completes the work reliably. Conversely, a lower-cost model can improve the economics of high-volume processes where human review remains manageable.
Potential areas to assess include:
- Extracting and interpreting information from forms and visual documents.
- Matching product or catalog information where image and text context matter.
- Preparing proposal drafts from supplied material.
- Supporting computer-use workflows that require precise interaction steps.
The availability in ChatGPT Work and Codex gives teams two distinct ways to explore the model. ChatGPT Work and Codex can support hands-on evaluation in their respective environments, while the API enables integration into an existing application or automated process. The appropriate route depends on whether the goal is individual productivity, development work or a repeatable operational workflow.
How to evaluate potential ROI
A useful pilot begins with a bounded task that has a known volume and a clear quality threshold. For example, a team can measure how long a person currently spends interpreting a form or preparing a first proposal draft, then compare that baseline with a Terra-assisted process. Include review time, exceptions and rework, not just model token charges.
Track a small set of practical measures:
- Completion time per task.
- Output acceptance rate before material edits.
- Human review and correction time.
- API token use and total cost per completed task.
- Error patterns involving visual inputs or computer actions.
This approach turns OpenAI's lower-cost positioning into an evidence-based decision. It also helps teams identify whether the model is best used for first drafts, structured extraction, human-in-the-loop assistance or a more automated workflow.
OpenAI also highlights safety and alignment improvements and provides a System Card covering deployment safeguards. That context matters for any implementation involving important customer, financial or operational information. Safeguards do not remove the need to define what the model may do, what requires review and how errors will be handled in a specific workflow.
Migration considerations for existing Terra users
Teams already using GPT-5.6 Terra should avoid treating a model change as a pricing-only update. The announced performance comparison suggests a strong reason to test GPT-6.1 Terra, but output style, visual interpretation behavior and tool interactions can affect downstream systems.
Start by replaying representative inputs from the current workflow and comparing outputs against the same acceptance criteria. Check prompts, structured output handling, any computer-use steps and the quality of exceptions. If the process has customer-facing or irreversible actions, retain human approval during the initial evaluation.
For API users, the confirmed implementation detail is straightforward: the new model identifier is gpt-6.1-terra. The harder part is operational validation. A staged pilot can reveal whether reduced token prices and OpenAI's claimed per-task efficiency translate into savings after review, software integration and process changes are included.
Businesses that want to move beyond ad hoc testing need a workflow that connects model outputs to real tools, approval steps and exception handling. Scalevise can help map a suitable process, test the economics and build AI workflow automation for operational tasks without forcing critical work into an unproven fully automated path. The immediate opportunity is to identify one repeatable, measurable task where visual reasoning or computer use can reduce manual effort. Discuss an AI automation project with Scalevise.
Frequently Asked Questions
What is GPT-6.1 Terra?
GPT-6.1 Terra is an OpenAI model designed around enhanced visual reasoning, precise computer-use capabilities and clearer writing. OpenAI positions it as a lower-cost option for relevant tasks.
Where is GPT-6.1 Terra available?
OpenAI says GPT-6.1 Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex. Developers can access it through the OpenAI API as gpt-6.1-terra.
How much does the GPT-6.1 Terra API cost?
Standard API pricing is $1 per million input tokens, $0.05 per million cached input tokens and $5 per million output tokens.
Should teams migrate from GPT-5.6 Terra immediately?
Teams should test GPT-6.1 Terra on representative workloads before migrating critical processes. Compare output quality, review time, tool behavior and total cost per completed task.
Conclusion
GPT-6.1 Terra extends OpenAI's lower-cost positioning for visual reasoning and computer-use work across ChatGPT Work, Codex and the API. Its published pricing and broad access create a credible reason to evaluate relevant workflows, but the business case depends on measured task quality, review requirements and real end-to-end costs.