OpenAI
OpenAI's GPT-6.1 Terra halves input prices from GPT-5.6 Terra
OpenAI's GPT-6.1 Terra is released in the API at $1 per 1M input tokens and $5 per 1M output, with a 1.05M-token context window.
OpenAI released GPT-6.1 Terra, a model that charges half as much for input as GPT-5.6 Terra and about 58% less for output. Developers running image-heavy or computer-use workloads on the older model can lower their bills, provided the new one performs as well on their tasks.
Terra Is Released in the API
OpenAI added the model to its API changelog on Oct. 26, 2026, calling it “a fast, efficient model for visual reasoning, precise computer use, and clear, natural writing.” The model ID is gpt-6.1-terra. It runs on the Responses API and Chat Completions. The entry lists it as released, and the model page lists standard pricing.
OpenAI’s GPT-6 guide lists the family as GPT-6 Astra, GPT-6.1 Sol, GPT-6.1 Terra and GPT-6 Luna.
Pricing Starts at $1 In and $5 Out
GPT-6.1 Terra costs $1 per 1M input tokens and $5 per 1M output tokens, according to OpenAI’s model page, read Oct. 26. GPT-5.6 Terra is listed at $2 for input and $12 for output. GPT-6.1 Sol is listed at $2 and $10, against Terra’s $1 and $5.
Cached input, which is prompt text the model has already processed, costs $0.05 per 1M tokens, down from $0.20 on GPT-5.6 Terra. Cache writes cost $1.25 with 30-minute retention and $2.50 with 12-hour retention.
A prompt over 272,000 input tokens is priced at 2x the input and cache rates and 1.5x the output rate for the full request. Batch and Flex processing are 50% lower. Regional processing adds 10%. Fast mode costs 2x Standard.
The Context Window Runs to 1.05M Tokens
The context window is 1,050,000 tokens, with a maximum input of 922,000 and a maximum output of 128,000. The model accepts text and images and returns text only. Its knowledge cutoff is June 1, 2026.
Reasoning effort, the setting that controls how much the model works through a problem before answering, accepts none, low, medium, high, xhigh and max. Medium is the default. GPT-6 Astra and GPT-6.1 Sol do not offer none.
The Batch endpoint is supported. The model does not support fine-tuning, Realtime, embeddings or image generation.
Tool Calls Must Move to the Responses API
Tool calling on GPT-6.1 Terra requires the Responses API. OpenAI’s GPT-6 guide, read Oct. 26, tells developers whose apps call tools through Chat Completions to migrate those requests.
The guide tells anyone moving from GPT-5.6 Terra to compare both models on representative images and workflows at the same reasoning effort. It suggests Terra for matching product photos to catalog entries, comparing technical drawings, acting on application screens and drafting or revising text. Fast mode for Terra supports EU data residency.
Our View
We think the migration rule matters most. A team on GPT-5.6 Terra that sends tool calls through Chat Completions cannot swap the model ID and stop. GPT-6.1 Terra takes tool calls only through the Responses API. That team rewrites its requests first, and the lower price applies after.
The rewrite is a cost paid once. The price difference repeats on every request the team sends.
The guide’s advice to test at the same reasoning effort keeps that comparison fair. A team that changes the effort setting and the model together cannot tell which one changed its results. The savings hold only if GPT-6.1 Terra matches GPT-5.6 Terra on that team’s own images and workflows.