Skip to content

AI Compass / News

GPT-6.1 Terra: OpenAI model at $1 input and $5 output per million tokens

General Costs

OpenAI has released GPT-6.1 Terra: what its list price says compared with Claude, and why you should test it yourself before switching.

Source: Changelog: GPT-6.1 Terra (OpenAI API Docs),

Note: I use Claude and Claude Code for my own work. Anthropic does not pay me, and the links are not affiliate links.

Technical drawing: a bar chart in US dollars per million tokens with one bar each for input and output. Haiku 5.5 sits at 0.10 / 0.50, the highlighted GPT-6.1 Terra at 1 / 5, Sonnet 5.5 at 2 / 10.

What happened: On 26 October 2026, OpenAI released GPT-6.1 Terra (gpt-6.1-terra), which it describes as a fast, efficient model for visual reasoning, precise computer use and natural writing. The model page lists 1 US dollar per million input tokens and 5 US dollars per million output tokens, 1 / 5 for short. Tokens are the chunks of text that usage is billed by. The context window, the amount of text per request, is 1.05 million tokens. Above 272,000 input tokens the whole request costs more: twice the input price, 1.5 times the output price. Batch and Flex cost half. Terra is available through the API, with tool calls only via the Responses API.

What this means: In OpenAI’s price list, Terra sits between GPT-6 Luna (0.10 / 0.50) and GPT-6.1 Sol (2 / 10). For comparison, Anthropic’s pricing page lists Claude Sonnet 5.5 at 2 / 10 and Claude Haiku 5.5 at 0.10 / 0.50, or 0.50 / 2.50 for prompts over 100,000 tokens. That adds another mid-priced option for automated workflows. But the price says nothing about quality on your task. Nor is it directly comparable: according to Anthropic, how many tokens the same text produces depends on the model’s tokenizer, the language and the content.

What you can do:

  1. Measure cost: Send 20 to 50 typical requests from your workflow to both models and compare the tokens you are billed for, not the price list.
  2. Check quality blind: Rate the outputs against the same criteria without knowing which model produced which.
  3. Watch long inputs: If your prompts often exceed 272,000 tokens (Terra) or 100,000 tokens (Haiku 5.5), higher prices apply.
  4. Count the switching effort: Prompts, tests and integrations take time, even when the token price drops.

Context: Terra’s strengths are vendor claims. All prices in US dollars per million tokens, as of 27 October 2026.

Bringing AI into your own team? That’s what my workshops are for. See the workshops