Wain AI/Tech Blog

AI news and trends worldwide, updated nearly every day

OpenAI Releases GPT-6.1 Terra in the API - $1 Input and $5 Output, Half the Price of GPT-6.1 Sol

OpenAI Releases GPT-6.1 Terra in the API - $1 Input and $5 Output, Half the Price of GPT-6.1 Sol

On October 26, 2026, OpenAI released GPT-6.1 Terra in its API, a model aimed at visual reasoning and computer use. It costs $1 per million input tokens and $5 per million output tokens, half the price of GPT-6.1 Sol. General availability in GitHub Copilot began the same day. A look at pricing, specifications, and what to watch when migrating.

On October 26, 2026, OpenAI used its API changelog to announce the release of GPT-6.1 Terra 1. The company describes it as a quick, efficient model built for visual reasoning, accurate operation of computers, and clear, natural prose. The model ID is gpt-6.1-terra, and it can be called through both the Responses API and the Chat Completions API.

Standard pricing per million tokens is $1 for input, $0.05 for cached input, and $5 for output 2. Compared with GPT-6.1 Sol, every one of those three rates is half as much. GitHub announced general availability in Copilot on the same day 4.

Half the Price of GPT-6.1 Sol

Here are the standard rates (per million tokens) as listed on each model’s page 256.

ModelInputCached inputOutput
GPT-6.1 Terra$1$0.05$5
GPT-5.6 Terra$2$0.2$12
GPT-6.1 Sol$2$0.1$10

GPT-5.6 Terra reached $2 input and $12 output through the July 30 price cut. GPT-6.1 Sol’s standard processing costs $2 for input and $10 for output.

The GPT-6.1 Terra model page also spells out the finer conditions 2. Writing to the cache costs $1.25 with 30-minute retention and $2.50 with the 12-hour retention that became available on October 20. Once a prompt goes past 272K input tokens, the whole request is billed with input and cache prices doubled and the output price raised by half. Batch and Flex are half of Standard, Fast mode is twice Standard, and regional processing carries a 10% surcharge.

Specifications That Matter for API Use

Pulling the implementation-relevant points from the model page and the GPT-6 guide 23:

  • Reasoning effort (reasoning.effort) has six levels — none, low, medium, high, xhigh, and max — with medium as the default. GPT-6 Astra and GPT-6.1 Sol lack the none level; Terra has it
  • Calling tools requires the Responses API. The guide asks apps that call tools through Chat Completions to move those requests over to the Responses API
  • It takes text and images as input and returns text; audio and video are not supported. The context window is 1,050,000 tokens (with up to 922,000 of input), output tops out at 128,000 tokens, and the knowledge cutoff is June 1, 2026
  • Tools available through the Responses API include computer_use, mcp, hosted_shell, and apply_patch, among others. Fine-tuning is not supported

As example uses, the model page lists pairing product images with catalog listings, reading technical drawings, operating on application screens, and drafting and revising text 2. For anyone moving over from GPT-5.6 Terra, the guide recommends testing both models against typical images and workflows of your own, with reasoning effort held constant 3. It also lays out a procedure: begin at medium, and raise the effort only if the quality gain justifies the extra time and tokens.

On by Default in Copilot Business and Enterprise

According to GitHub, GPT-6.1 Terra is available on Copilot Pro, Pro+, Max, Business, and Enterprise, and can be picked from the model picker in VS Code, Visual Studio, Copilot CLI, the cloud agent, github.com, JetBrains IDEs, and elsewhere 4. The rollout is gradual. GitHub pitches it as cheaper than GPT-6.1 Sol and suited to routine agent-driven coding and command-line work. Under usage-based billing, it is charged at the provider’s list price.

What organization administrators should keep an eye on is how it gets turned on. According to the changelog, when default model enablement is in effect a newly added model turns on by itself, and it stays off only where an administrator has disabled the global default or blocked this particular model 4. As noted in our article on the default policy for new features, Copilot runs its models-side default policy separately from the features side. If nobody has changed that default, Terra will reach users as is.

What Half of Sol’s Price Means

With both input and output at half of GPT-6.1 Sol’s rates, it may now be worth re-examining which model to use for workloads that read images or operate screens at volume. How large the quality gap is, though, depends on the use case and cannot be judged from the price list alone.

The reliable way to decide whether to switch is the one the guide recommends: compare results and cost on your own typical tasks with reasoning effort held equal. For apps that use tools through Chat Completions, moving to the Responses API comes as part of the same change.

Sources

  1. Changelog - OpenAI API documentation (entry for October 26, 2026)
  2. GPT-6.1 Terra - OpenAI API documentation (model page)
  3. Using GPT-6 - OpenAI API documentation (GPT-6 guide)
  4. GPT-6.1 Terra in GitHub Copilot - GitHub Changelog (October 26, 2026)
  5. GPT-5.6 Terra - OpenAI API documentation (model page)
  6. GPT-6.1 Sol - OpenAI API documentation (model page)

We publish the latest AI news nearly every day.

Subscribe via RSS Get new posts the moment they go live.

Search other keywords →