GPT-6.1 Terra arrived on Bedrock and in Copilot the same day, with different pitches
Serguey Asael Shinder1 min readOpenAI's GPT-6.1 Terra became generally available in two places other than OpenAI's own products on 26 October: Amazon Bedrock and GitHub Copilot. Read side by side, the two announcements show how differently the same model is being positioned.
AWS's What's New post (https://aws.amazon.com/about-aws/whats-new/2026/10/openai-gpt-6-1-terra-on-amazon-bedrock) leads with economics and agents. It says Terra scored 15 percentage points higher than GPT-5.6 Terra on OpenAI's Agents' Last Exam evaluation, performed on par with GPT-6 Astra, completed tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol, and has input and output prices at least 50% lower than GPT-6.1 Sol and GPT-5.6 Terra. It supports up to 1 million input tokens and accepts text and images, and AWS notes that ChatGPT Work and Codex can be configured to use the model on Bedrock.
GitHub's changelog (https://github.blog/changelog/2026-10-26-gpt-6-1-terra-in-github-copilot) makes a narrower claim: Terra is for everyday agentic coding and terminal work at lower cost than GPT-6.1 Sol, and in early testing it solved noticeably more terminal and C# tasks than GPT-5.6 Terra. It is billed at provider list price under usage-based billing, available on Pro, Pro+, Max, Business and Enterprise, and — unless an administrator has turned off default enablement — switched on automatically.
What it means
Two things are worth separating. The benchmark figures are OpenAI's own evaluation, repeated by AWS; neither announcement publishes independent numbers. And the default-on rule in Copilot means a new model with its own price can appear in an organisation's model picker without anyone choosing it. For teams on usage-based billing, the first practical step is not a benchmark but a look at the model policy page.