GPT-6.1 Terra API: $1/$5 Pricing, Benchmarks and Setup

GPT-6.1 Terra is OpenAI's new fast, low-cost model, released on 26 October 2026 as gpt-6.1-terra. GPT-6.1 Terra API pricing is $1 per 1M input tokens, $0.05 per 1M cached input tokens, and $5 per 1M output tokens. That is half the input price of GPT-6.1 Sol and half of GPT-5.6 Terra. The context window is 1,050,000 tokens.

GPT-6.1 Terra is not on RouterPlex yet. Register for a RouterPlex account now and your key will be ready the day the catalog lists gpt-6.1-terra. Until then, the same key calls GPT-6.1 Sol and GPT-5.6 Terra at vendor list price.

GPT-6.1 Terra API pricing #

OpenAI's model card lists these standard rates, in USD per 1M tokens.

Cached input (the part of a prompt the provider has seen recently) costs 5% of the normal input rate. Batch and Flex are 50% below Standard. Fast mode is 2x Standard. Regional processing adds 10% where it is offered.

Prompts above 272,000 input tokens cost 2x input and cache rates and 1.5x output for the whole request. So a long prompt bills $2 input, $0.10 cached, and $7.50 output per 1M. The higher rate covers every token in that request, not only the tokens past the line.

A 40,000-token prompt that returns 2,000 tokens costs $0.05 on Terra with no cache. Reasoning tokens bill as output, so a high effort setting raises the output side.

GPT-6.1 Terra vs GPT-5.6 Terra, Sol, Luna, and Astra #

Model IDs and list prices per 1M tokens. The last column is the cost of one 40,000-in, 2,000-out turn with no cache.

The new model replaces GPT-5.6 Terra at less than half the cost per turn. The input rate drops from $2 to $1. Output drops from $12 to $5. The launch post puts Terra's standard API prices 50% lower than both GPT-5.6 Terra and GPT-6.1 Sol.

Sol is still the stronger coder. OpenAI's own post says GPT-6.1 Sol and GPT-6 Astra "remain the better choice" for the most complex coding tasks. GPT-6 Luna is ten times cheaper again for bulk extraction.

GPT-6.1 Terra benchmarks for computer use #

OpenAI tested computer use on Agents' Last Exam, a set of professional tasks such as turning architectural drawings into a 3D model. At medium reasoning effort, GPT-6.1 Terra scores 15 percentage points above GPT-5.6 Terra at about one third of the cost. In OpenAI's results it also performs on par with GPT-6 Astra at one tenth of Astra's price per token.

At max effort, OpenAI reports Terra uses about one third of the tokens Claude Opus 5.5 uses. Combined with one fourth of the token price, that gives roughly 9x lower cost per task. OpenAI also says Terra finishes tasks over three times faster than Astra and GPT-6.1 Sol.

On OSWorld 2.1, which tests agents using desktop apps, Terra at medium effort beats GPT-5.6 Terra's best score at a fraction of the cost per task.

GPT-6.1 Terra vision: MMMU-Pro #

Terra accepts text and image input and returns text. On MMMU-Pro, a visual reasoning test, OpenAI says GPT-6.1 Terra beats GPT-5.6 Terra at every reasoning setting. It nearly matches GPT-6 Astra and GPT-6.1 Sol at a much lower cost.

That fits the jobs OpenAI names for it. Examples include matching product photos to catalog entries, reading technical drawings, and copying invoice details into accounting software.

GPT-6.1 Terra coding and factual accuracy #

On Terminal-Bench 4.0, a test of command-line coding agents, GPT-6.1 Terra solves 50.3% of tasks at max effort. GPT-5.6 Terra's published score was 21.5%, so the new model more than doubles it.

On Terminal-Bench-Science 0.1, Terra at max effort beats GPT-6.1 Sol at lower cost. That run lands within five points of Opus 5.5 at less than one fifth of the estimated cost.

Hallucinations dropped sharply. On OpenAI's internal set of error-inducing prompts, the error rate at low effort fell from 53.6% on GPT-5.6 Terra to 16.7% on GPT-6.1 Terra. At max effort it fell to 4.3%, close to Astra and Sol.

These are vendor-run results. RouterPlex has not re-run them. Test your own tasks before you switch a production route.

GPT-6.1 Terra context window, effort levels, and limits #

The model card lists these specs, checked 26 October 2026:

• 1,050,000-token context window and 128,000 max output tokens.
• Knowledge cutoff of 1 June 2026.
• reasoning.effort accepts none, low, medium (the default), high, xhigh, and max.
• Tool calling needs the Responses API. Chat Completions works for plain text and image requests.
• Responses API tools include web search, file search, code interpreter, computer use, and MCP.
• The lowest paid API tier allows 5,000 requests and 1,000,000 tokens per minute. The free tier is not supported.

In ChatGPT, Terra is available to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex. The same plan allowance covers more tasks on Terra than on GPT-5.6 Terra.

How to use GPT-6.1 Terra on RouterPlex #

Create your RouterPlex account today. When the catalog lists gpt-6.1-terra, you change one string in your code. Until then, this request runs on the closest live model, gpt-6.1-sol:

curl https://api.routerplex.com/v1/chat/completions \
  -H "Authorization: Bearer $ROUTERPLEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6.1-sol",
    "messages": [
      {"role": "user", "content": "Read this invoice text and return vendor, date and total as JSON."}
    ]
  }'

The base URL is https://api.routerplex.com/v1 for OpenAI SDKs. Claude Code and the Anthropic SDK use https://api.routerplex.com. See What is an OpenAI-compatible API for SDK setup.

Every RouterPlex key works on one prepaid balance. Token prices are the vendor list price with 0% markup. Top-ups have a $0 fee and a $5 minimum. The balance stops at $0 and never goes negative. Each key can carry a hard spend limit. A $1 trial needs no card and does not expire.

Where it goes wrong #

• model_not_found for gpt-6.1-terra. The model is not in the RouterPlex catalog yet. Use gpt-6.1-sol or gpt-5.6-terra until the OpenAI hub shows Terra.
• Tool calls fail on Chat Completions. OpenAI's model card puts tool calling on the Responses API.
• A bill that doubled. Your prompt crossed 272,000 input tokens, so the whole request moved to the long-context rate.
• 402 at $0. Your prepaid balance ran out. Top up from your dashboard, or raise the key's budget. The pricing page lists the top-up terms.

Next step #

Register on RouterPlex so your key is ready for gpt-6.1-terra. Compare today's live OpenAI prices on the OpenAI model hub.