None
None
Long Story Video Skill
Turn your idea into 10s–10min video—AI writes scripts, prompts, and generates footage automatically.
Ads Video Skill
Generate professional ads and sales videos—AI auto-generates scripts, prompts, and footage.
3D Science Explainer Video Skill
Convert scientific concepts into stunning 3D explain animations
Feedback
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
GPT-6.1 Terra
GPT-6.1 Terra brings image-heavy reasoning, hands-on app control, and quick clip edits into one workflow, with API rates far under GPT-5.6 Terra.
All Tools
Discover our comprehensive AI-powered animation toolkit
Inside GPT-6.1 Terra: Reasoning That Also Acts
Detailed image understanding and exact on-screen control arrive in a single lean model, trimming the number of passes needed to finish a clip or a desktop job.
- Reads Images, Then ActsGPT-6.1 Terra interprets technical drawings and animation frames, then drives the applications required to turn that reading into finished work.
- Lower Cost per TokenStandard token rates land 50% under GPT-5.6 Terra and GPT-6.1 Sol, which keeps bulk invoice parsing and data entry affordable.
- Stretch Every PlanInside ChatGPT work and Codex, one allowance covers more jobs than GPT-5.6 Terra, and far more than GPT-6 Astra.
Running GPT-6.1 Terra in Three Moves
From a loose brief to a delivered file, the whole loop takes three moves.
GPT-6.1 Terra Test Scores, Explained
Desktop control, image understanding, factual reliability, and coding ability, tracked across Agents' Last Exam, MMMU-Pro, OSWorld 2.1, and Terminal-Bench.
Agents' Last Exam
On computer-use tasks at medium reasoning, GPT-6.1 Terra lands 15 points higher than GPT-5.6 Terra while spending roughly a third as much.
MMMU-Pro
Every reasoning level beats GPT-5.6 Terra, and it closes in on GPT-6 Astra and GPT-6.1 Sol at a fraction of the spend.
OSWorld 2.1
Even at medium reasoning it tops GPT-5.6 Terra's best result on long desktop chains, such as assembling a 3D model from drawings.
Factual Accuracy
On tricky prompts the error rate drops from 53.6% to 16.7% at low reasoning, then to 4.3% once reasoning is maxed out.
Terminal-Bench 4.0
With reasoning maxed it clears 50.3% of command-line tasks — more than twice GPT-5.6 Terra's 21.5%.
Terminal-Bench-Science 0.1
It edges out GPT-6.1 Sol while costing less, finishing within five points of Claude Opus 5.5 on science tasks driven by code.
GPT-6.1 Terra: Your Questions Answered
Straight answers on cost, where to access it, error rates, and how it handles desktop work.
What exactly is GPT-6.1 Terra?
A lean, quick model tuned for fine-grained image reasoning, accurate on-screen operation, and cleaner, more natural prose.
How much does the API cost?
List pricing runs $1 per million input tokens, $0.05 per million cached input tokens, and $5 per million output tokens — roughly half of GPT-5.6 Terra and GPT-6.1 Sol.
Where can I use GPT-6.1 Terra?
Plus, Pro, Enterprise, and Edu subscribers get it inside ChatGPT work and Codex, while developers reach gpt-6.1-terra through the OpenAI API.
How does it compare with GPT-5.6 Terra?
Medium reasoning adds 15 points on Agents' Last Exam, the API price halves, and jobs wrap up more than three times faster than GPT-6 Astra and GPT-6.1 Sol.
Does it make fewer factual errors?
It does. On internal prompts prone to error, the mistake rate falls from 53.6% to 16.7% at low reasoning, and to 4.3% at maximum.
How is it kept safe?
It clears the 'severe' cybersecurity bar in the Preparedness Framework, carries over GPT-6 Astra's safeguards, and holds up against prompt injection.
Put GPT-6.1 Terra to Work
Reach GPT-6.1 Terra through ChatGPT work, Codex, or the API — $1 per million input tokens in, $5 per million out.