I tried the new GPT-6.1 model "Terra" on Amazon Bedrock
This page has been translated by machine translation. View original
Introduction
OpenAI's GPT-6.1 Terra is now generally available on Amazon Bedrock.
GPT-6.1 Terra is the latest model in the Terra series, and AWS's announcement states that the input/output token price is half that of GPT-6.1 Sol.
GPT-6.1 Sol on Bedrock, which became GA on September 29, 2026, is introduced in the article below.
In this article, I tried calling GPT-6.1 Terra from Bedrock Runtime's Converse, Responses, and Chat Completions.
Model and Inference Profile
I ran get-foundation-model in us-east-1 to check Terra's model information.
aws bedrock get-foundation-model --region us-east-1 --model-identifier openai.gpt-6.1-terra
{
"modelDetails": {
"modelArn": "arn:aws:bedrock:us-east-1::foundation-model/openai.gpt-6.1-terra",
"modelId": "openai.gpt-6.1-terra",
"modelName": "GPT-6.1 Terra",
"providerName": "OpenAI",
"inputModalities": [
"TEXT",
"IMAGE"
],
"outputModalities": [
"TEXT"
],
"responseStreamingSupported": true,
"customizationsSupported": [],
"inferenceTypesSupported": [
"INFERENCE_PROFILE"
],
"modelLifecycle": {
"status": "ACTIVE",
"startOfLifeTime": "2026-10-23T10:00:00+00:00"
}
}
}
Since inferenceTypesSupported only includes INFERENCE_PROFILE, you need to specify an inference profile ID when making calls.
When I retrieved the list of system-defined inference profiles in us-east-1, Terra had two inference profiles: global.openai.gpt-6.1-terra and us.openai.gpt-6.1-terra. Both were ACTIVE. According to the profile descriptions, us. routes to three regions — us-east-1, us-east-2, and us-west-2 — while global. routes to all supported regions. This article uses global.openai.gpt-6.1-terra.
Calling from bedrock-runtime
I called Converse from the Tokyo region, specifying global.openai.gpt-6.1-terra.
aws bedrock-runtime converse \
--region ap-northeast-1 \
--model-id global.openai.gpt-6.1-terra \
--messages '[{"role":"user","content":[{"text":"Say ok"}]}]'
{
"output": {
"message": {
"role": "assistant",
"content": [
{
"text": "ok"
}
]
}
},
"stopReason": "end_turn",
"usage": {
"inputTokens": 8,
"outputTokens": 5,
"totalTokens": 13,
"cacheReadInputTokens": 0,
"cacheWriteInputTokens": 0
},
"metrics": {
"latencyMs": 959
}
}
I also called the OpenAI-compatible Responses API using curl with SigV4 signing.
curl --silent --show-error \
--aws-sigv4 'aws:amz:us-east-1:bedrock' \
--user "$AWS_ACCESS_KEY_ID:$AWS_SECRET_ACCESS_KEY" \
-H "x-amz-security-token: $AWS_SESSION_TOKEN" \
-H 'Content-Type: application/json' \
-d '{"model":"global.openai.gpt-6.1-terra","input":"Say ok","store":false}' \
-w '\nHTTP %{http_code}\n' \
https://bedrock-runtime.us-east-1.amazonaws.com/openai/v1/responses
Responses returned HTTP 200. Chat Completions also returned HTTP 200 when sent to /openai/v1/chat/completions with the same model ID.
Pricing
The table below shows OpenAI API Standard pricing. All prices are in USD per 1 million tokens, for input up to 272K tokens. Cache writes pricing applies to a 30-minute retention period.
| Model | Input | Cached input | Cache writes | Output |
|---|---|---|---|---|
| GPT-6.1 Sol | $2.00 | $0.10 | $2.50 | $10.00 |
| GPT-6.1 Terra | $1.00 | $0.05 | $1.25 | $5.00 |
For Bedrock pricing, in the case of GPT-5.6 Terra, the global inference profile price was the same as the price listed by OpenAI.
At the time of writing, the model card for GPT-6.1 Terra has not yet been published, so please check the documentation after it is updated for official pricing.
Summary
In OpenAI's GPT-6.1 series, following Sol released at the end of September, Terra is now also available on Bedrock.
Since it shares the same API specification as Sol, you can try it out simply by switching the model ID. Give it a try as a cost-effective model option.
