AI and modelsConfirmedOpenAI's GPT-6.1 Terra, a new AI model with enhanced visual reasoning and speed, is now generally available on Amazon Bedrock, offering a cost-effective solution for complex enterprise AI workloads.
Today, October 26, 2026, AWS announced the general availability of OpenAI GPT-6.1 Terra on Amazon Bedrock. This new model expands the GPT-6 family, designed to balance capability, speed, and cost efficiency for everyday workloads. GPT-6.1 Terra improves upon GPT-5.6 Terra by offering detailed visual reasoning, accurate computer use, and enhanced factuality and natural writing.
In OpenAI’s Agents’ Last Exam evaluation, GPT-6.1 Terra scored 15 percentage points higher than GPT-5.6 Terra, performed comparably to GPT-6 Astra, and completed tasks more than three times faster than both GPT-6 Astra and GPT-6.1 Sol. The model provides significant cost savings, with input and output token prices at least 50% lower than GPT-6.1 Sol and GPT-5.6 Terra. It supports a substantial context window of up to 1 million input tokens and accepts both text and image inputs.
Amazon Bedrock's inference engine provides the necessary performance, security, and scale for enterprise workloads. Developers can access GPT-6.1 Terra directly via Amazon Bedrock APIs or configure existing tools like ChatGPT Work and Codex to utilize the model on Bedrock. Practical applications include drafting proposals, interpreting technical drawings, investigating codebases, and building agents that interact with business systems. AWS controls are available for workload security, access governance, and auditing model invocation.
The general availability of OpenAI GPT-6.1 Terra on Amazon Bedrock represents a notable advancement for software engineers and enterprises leveraging cloud-based AI. The model's confirmed capabilities, particularly its multimodal input, large context window, and performance metrics, offer direct implications for developing and deploying sophisticated AI applications.
GPT-6.1 Terra's confirmed improvements in detailed visual reasoning, accurate computer use, and enhanced factuality directly address common challenges in enterprise AI adoption. The ability to process and interpret images alongside text inputs, for instance, enables new categories of applications that require understanding visual data, such as automated analysis of technical diagrams, medical imagery, or user interface screenshots. This multimodal capability, combined with "accurate computer use," indicates the model's suitability for agentic workflows where AI systems need to interact with digital environments, interpret visual cues, and execute actions based on complex instructions. The 15 percentage point increase over GPT-5.6 Terra in OpenAI’s Agents’ Last Exam evaluation provides a quantifiable measure of this enhanced agentic performance. For developers, this means the potential to build more robust and versatile AI agents capable of handling tasks that span both textual and visual domains, reducing the need for separate, specialized models or complex pre-processing pipelines. The improved factuality and natural writing also contribute to higher quality outputs, reducing the need for extensive post-processing or human review in content generation and summarization tasks.
The operational characteristics of GPT-6.1 Terra, specifically its speed and cost efficiency, are critical for enterprise-scale deployments. The confirmed speed improvement, completing tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol, directly translates to lower latency for real-time applications and higher throughput for batch processing. This performance boost, coupled with input and output token prices at least 50% lower than previous models, significantly reduces the total cost of ownership for AI-powered solutions. For developers, this economic advantage makes it feasible to deploy AI models in scenarios where cost was previously a prohibitive factor, enabling more frequent invocations or broader application across an organization. Furthermore, the support for a context window of up to 1 million input tokens is a substantial engineering enabler. This large context allows developers to feed entire documents, extensive codebases, or prolonged conversation histories into the model in a single request. This capability can simplify prompt engineering, reduce the complexity of retrieval-augmented generation (RAG) architectures for certain use cases, and improve the model's ability to maintain coherence and context over long interactions. It allows for more comprehensive analysis without the overhead of managing external memory or chunking strategies.
The general availability of GPT-6.1 Terra on Amazon Bedrock provides a streamlined path for enterprise adoption, leveraging AWS's established infrastructure. Bedrock's inference engine is confirmed to deliver performance, security, and scale, which are non-negotiable requirements for production workloads. Developers benefit from AWS's existing controls for workload security, access governance, and auditing model invocation activity, reducing the burden of implementing these critical features independently. The flexibility to call GPT-6.1 Terra directly through Amazon Bedrock APIs or configure existing tools like ChatGPT Work and Codex offers diverse integration options. This means developers can either build custom applications from scratch using the Bedrock API or upgrade existing OpenAI-based workflows to leverage Terra's enhanced capabilities without extensive refactoring. The confirmed use cases—drafting proposals, interpreting technical drawings, investigating codebases, and building agents that operate across business systems—directly illustrate how the model's multimodal, large-context, and agentic capabilities can be applied to solve real-world enterprise problems. For instance, investigating codebases benefits directly from the large context window to analyze entire repositories and the "accurate computer use" for understanding code logic.