OpenAI has launched GPT-6.1 Terra, a model designed for visual reasoning and computer use, replacing GPT-5.6 Terra. It is now live on the API under the identifier gpt-6.1-terra with standard pricing of $1 per million input tokens and $5 per million output tokens.
Key specifications and availability
The model accepts text and image input, returning text responses. Its context window holds 1,050,000 tokens, with a maximum output of 128,000 tokens. Knowledge cutoff stands at June 1, 2026.
Reasoning effort ranges from none to max, defaulting to medium. Tool calling requires the Responses API. Supported tools include computer use, hosted shell, code interpreter, web search, file search, and MCP. OpenAI lists matching product photos to catalog entries and interpreting technical drawings as primary use cases.
Access is restricted to OpenAI-hosted environments including the API, ChatGPT Work, and Codex. Weights remain closed.
Performance on computer use tasks
OpenAI cites Agents’ Last Exam v1 as a primary test, featuring professional computer tasks such as converting architectural drawings into 3D models.
At medium effort, Terra scores 57.6% at a cost of $0.29 per task. GPT-5.6 Terra achieved 42.6% at $0.84 per task under the same conditions. GPT-6 Astra matches the 57.6% score but costs $4.82 per task.
At max effort, Terra reaches 60.0% at $1.01 per task. Claude Opus 5.5 scores 63.2% at $8.82 per task on the same chart. OpenAI states Terra completes tasks over three times faster than GPT-6 Astra and GPT-6.1 Sol.
On the OSWorld 2.1 offline set, Terra scores 66.7% at medium effort and 74.2% at max. GPT-5.6 Terra peaked at 57.1%. GPT-6.1 Sol reaches 78.0%, GPT-6 Astra 80.9%, and Claude Sonnet 5.5 83.9%.
Visual reasoning capabilities
On MMMU-Pro (Vision), Terra scores 85.5% at max effort. GPT-6.1 Sol scores 86.3% and GPT-6 Astra 86.7%. GPT-5.6 Terra topped out at 81.7%. Terra’s max-effort run costs $0.012 per task, against $0.065 for Astra.
Factuality and coding results
On OpenAI’s internal set of error-inducing prompts, the low-effort error rate fell from 53.6% to 16.7%. At max effort, it fell 18 points to 4.3%. The set is adversarial by design, so production error rates will differ.
On Terminal-Bench 4.0, Terra solves 50.3% at max effort, up from 21.5% for GPT-5.6 Terra. It still trails GPT-6.1 Sol (55.9%), GPT-6 Astra (58.2%) and Claude Sonnet 5.5 (70.6%). OpenAI itself recommends Sol or Astra for the most complex coding tasks.
On Terminal-Bench-Science 0.1, Terra scores 58.3% at $4.15 per task. That beats GPT-6.1 Sol’s 57.0% at $5.47. Claude Opus 5.5 with fallbacks scores 63.3% at $23.21.
Pricing structure and access
Standard API pricing is $1 input, $0.05 cached input and $5 output per million tokens. Batch and Flex options are 50% cheaper. Fast mode costs 2x Standard. Prompts above 272K input tokens bill at 2x input and 1.5x output.
This represents half of GPT-6.1 Sol’s $2/$10 rate and one-tenth of GPT-6 Astra’s $10/$50 rate. Claude Opus 5.5 lists at $4/$20.
In ChatGPT, Terra is live for Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex.
Comparison with competitors
The following table compares key metrics across models. Note that benchmark scores and per-task costs are read from OpenAI’s launch charts and are self-reported. Specs and prices come from each vendor’s model page.
Model comparison table
- Parameters: Not disclosed for all models listed.
- Context window: 1.05M for Terra, Sol, and Astra; 1M for Claude Opus 5.5 and Sonnet 5.5.
- Max output: 128K for all models.
- Knowledge cutoff: Jun 1, 2026 for Terra; Apr 30, 2026 for Sol and Astra; Jun 2026 for Claude models.
- API price (in/out per 1M): Terra ($1/$5); Sol ($2/$10); Astra ($10/$50); Claude Opus 5.5 ($4/$20); Claude Sonnet 5.5 ($2/$10).
- Weights: Closed for all models.
- Agents’ Last Exam v1 (max): Terra 60.0% ($1.01/task); Sol 58.6% ($2.00/task); Astra 59.3% ($7.24/task); Opus 5.5 63.2% ($8.82/task); Sonnet 5.5 not shown.
- MMMU-Pro Vision (max): Terra 85.5%; Sol 86.3%; Astra 86.7%; Opus 5.5 and Sonnet 5.5 not shown.
- OSWorld 2.1 offline (max): Terra 74.2%; Sol 78.0%; Astra 80.9%; Opus 5.5 not shown; Sonnet 5.5 83.9%.
- Terminal-Bench 4.0 (max): Terra 50.3%; Sol 55.9%; Astra 58.2%; Opus 5.5 not shown; Sonnet 5.5 70.6%.
- Terminal-Bench-Science 0.1 (max): Terra 58.3%; Sol 57.0%; Astra 68.1%; Opus 5.5 63.3% (with fallbacks); Sonnet 5.5 not shown.
What it means
Teams automating screens, forms, and documents at volume now have a cheaper option. Terra matches GPT-6 Astra’s performance on Agents’ Last Exam but at a fraction of the cost. Visual reasoning scores remain high, within 1.2 points of Astra. Coding capabilities still lag behind competitors, particularly on Terminal-Bench 4.0 where Sonnet 5.5 outperforms Terra by 20 percentage points. All scores are OpenAI-reported; parameters and hardware remain undisclosed.
