Intelligence
Index v5.0
Overview
From visual understanding to action
GPT-6.1 Terra brings visual reasoning, computer use and writing into one model. It can interpret screenshots and technical documents, then use that context to work through a task. OpenAI positions it for everyday agent workflows, including document processing, form filling and software maintenance.
The release evaluates this balance on Agents’ Last Exam: professional tasks that require an agent to work across applications. The chart below shows task score against cost at different reasoning settings. Terra’s green curve makes it possible to see how additional reasoning changes both the result and the cost, alongside the other models in the evaluation.
Choose the model around the work
Terra is a useful candidate for recurring tasks where visual understanding and response cost matter together. For the hardest software engineering work, OpenAI still recommends evaluating Sol and Astra. The independent comparisons below add another perspective on intelligence, generation speed and benchmark task cost.
Benchmarks by 
Artificial Analysis
See how GPT-6.1 Terra compares in intelligence, generation speed and cost per task.
Benchmark results
Compare model performance across four independent benchmarks.
FrontierCode 2.0
Agentic software engineering in open-source repositories.
HLE-Diamond
A demanding evaluation of reasoning and knowledge.
AutomationBench-AA
Completing agentic workflows across SaaS applications.
Terminal-Bench-Science 0.1
Scientific research workflows performed by an agent in a terminal.
Pricing
| Usage | Price | Unit |
|---|---|---|
| Input | $1 | 1M tokens |
| Output | $5 | 1M tokens |
| Cached input | $0.05 | 1M tokens |
Compare token prices
Published input and output rates for five models, per million tokens.
| Model | Input | Output |
|---|---|---|
 GPT-6.1 Terra | $1.00 | $5.00 |
| GPT-6.1 Sol | $2.00 | $10.00 |
 Kimi K3.1 Pro | $2.00 | $10.00 |
 Claude Sonnet 5.5 | $2.00 | $10.00 |
| Claude Opus 5.5 | $4.00 | $20.00 |
Quick start
View docs (opens in a new tab)
Build with your agent
Give your agent the model details, code and documentation.
Give your agent a head startCopy this model’s API details, a working request and documentation links. Paste them into your coding agent to build the integration.
1Get your API key
Create API key (opens in a new tab)export AIMLAPI_KEY="8f2b4c7d9a1e6b3f…"2Make your first request
curl 'https://api.aimlapi.com/v1/chat/completions' \
-H "Authorization: Bearer $AIMLAPI_KEY" \
-H 'Content-Type: application/json' \
-d '{
"model": "openai/gpt-6.1-terra",
"messages": [
{
"role": "user",
"content": "Summarize three practical ways to improve an API onboarding flow."
}
]
}'import os
import requests
headers = {"Authorization": "Bearer " + os.environ["AIMLAPI_KEY"]}
payload = {
"model": "openai/gpt-6.1-terra",
"messages": [
{
"role": "user",
"content": "Summarize three practical ways to improve an API onboarding flow."
}
]
}
response = requests.post(
"https://api.aimlapi.com/v1/chat/completions",
headers=headers,
json=payload,
timeout=120,
)
response.raise_for_status()
data = response.json()
print(data["choices"][0]["message"]["content"])async function main() {
const headers = {
Authorization: "Bearer " + process.env.AIMLAPI_KEY,
"Content-Type": "application/json",
};
const response = await fetch(
"https://api.aimlapi.com/v1/chat/completions",
{
method: "POST",
headers,
body: JSON.stringify({
"model": "openai/gpt-6.1-terra",
"messages": [
{
"role": "user",
"content": "Summarize three practical ways to improve an API onboarding flow."
}
]
}),
},
);
if (!response.ok) throw new Error(await response.text());
const data = await response.json();
console.log(data.choices[0].message.content);
}
main().catch(error => {
console.error(error);
process.exitCode = 1;
});Related models
Related articles
Frequently asked questions
How much does GPT-6.1 Terra cost?
Standard pricing is $1 per million input tokens and $5 per million output tokens. Cached input costs $0.05 per million tokens.
What is cached input?
Cached input is the part of a prompt reused from the prompt cache. Eligible cached tokens cost $0.05 per million instead of $1, a 95% lower input rate. New input and generated output are billed at their regular rates.
Can GPT-6.1 Terra process images?
Yes. You can combine text and images in a request to ask questions about screenshots, diagrams or other visual content. The model returns text.
Does it support tool calling and structured output?
Yes. Parallel tool calls and structured output are supported through the Responses API.
What is the context window?
GPT-6.1 Terra has a 1.05-million-token context window for the prompt, conversation history and generated response.
What is the maximum output length?
The published maximum output is 128,000 tokens, within the available context window.