What are the input and output costs for GPT-6.1 Terra?

Standard input tokens cost $1.00 per million, cached input tokens cost $0.05 per million, and output tokens cost $5.00 per million.

What is the context window limit for GPT-6.1 Terra?

The model supports approximately 1,050,000 tokens in its context window and allows up to 128,000 output tokens per single response.

How does GPT-6.1 Terra perform on Terminal-Bench 4.0?

At max reasoning effort, GPT-6.1 Terra successfully solves 50.3% of tasks on Terminal-Bench 4.0.

What modalities does GPT-6.1 Terra support?

The API accepts image, text, and file inputs while outputting text only, and supports tool calling via the Responses API.

GPT-6.1 Terra Pricing and Specs

GPT-6.1 Terra API Price and Benchmarks. Macro lens view of a dense silicon interposer glowing with amber thermal conduction gel beneath a brushed aluminum water block, threaded with rigid black coolant lines carrying reflections of amber — GPUniq

Data as of October 27, 20263 sources

Drafted with AI tools and edited by a human. The figures and conclusions were checked by a GPUniq editor.

GPT-6.1 Terra Costs

The OpenAI API charges $1.00 per million standard input tokens for gpt-6.1-terra. Cached input tokens drop to $0.05 per million. Output tokens cost $5.00 per million.

Token Type	Price, USD per Million Tokens
Standard Input	1.00
Cached Input	0.05
Output	5.00

Context Window Limits

The model supports roughly 1,050,000 tokens in its context window. You can request up to 128,000 output tokens per single response.

Terminal-Bench 4.0 Performance

On Terminal-Bench 4.0, the model solves 50.3% of tasks at max reasoning effort. The predecessor GPT-5.6 Terra managed only 21.5% on the same benchmark.

Error Rates

Internal tests show an error rate of 4.3% on difficult prompts at max reasoning effort. Running at low effort yields a 16.7% error rate. GPT-5.6 Terra registered a 53.6% error rate at low effort.

Exam Scores

On Agents' Last Exam at medium reasoning effort, GPT-6.1 Terra scores about 15 points higher than GPT-5.6 Terra. This score matches GPT-6 Astra.

Supported Modalities

The API accepts image, text, and file inputs while outputting text only. It supports tool calling through the Responses API. You can access it via standard API endpoints like Chat Completions, Responses, and v1/live/sessions, as well as through ChatGPT Work and Codex.

Methodology

Facts are compiled from the vendor's own announcement and documentation, linked below, at the time of writing; where the vendor hasn't published a number it is marked as such. GPUniq prices, when the model is already available here, come from our live catalog.