OpenAI Launches GPT-6.1 Terra, Half the Price in the API

Key Takeaways

1API pricing: $1 per million tokens for input and $5 for output

2Access starting October 26 for Plus, Pro, Enterprise, and Edu in ChatGPT Work and Codex

3Execution over three times faster than Sol and Astra, according to OpenAI

Why it matters : OpenAI offers a model touted as faster and 50% cheaper in the API, with immediate availability for professional plans.

OpenAI deploys GPT-6.1 Terra to professional clients and via its API, with prices set at $1 per million tokens for input and $5 for output, which is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol. The model is not yet available in the public ChatGPT. The company highlights execution that is more than three times faster than Sol and Astra, improvements in factual errors, and a level of safeguards aligned with Astra.

Immediate access for professionals and API, announced prices of $1 and $5

OpenAI offers GPT-6.1 Terra at $1 per million tokens for input and $5 per million for output via its API. The company specifies that these prices are 50% lower than those of GPT-5.6 Terra and GPT-6.1 Sol. Since October 26, Plus, Pro, Enterprise, and Edu subscribers have been able to access GPT-6.1 Terra in ChatGPT Work and Codex, but the model is not yet available in ChatGPT. Developers also have access via the API under the identifier gpt-6.1-terra. The deployment plans to make it available in ChatGPT after ChatGPT Work and Codex.

Speed and accuracy: over 3x faster than Sol and Astra, fewer errors

According to OpenAI, GPT-6.1 Terra performs tasks at a speed more than three times that of GPT-6 Astra and GPT-6.1 Sol. In an internal benchmark of queries designed to induce errors, the error rate with low reasoning effort dropped from 53.6% with GPT-5.6 Terra to 16.7% with GPT-6.1 Terra. On Agents’ Last Exam, a benchmark of professional tasks on a computer, GPT-6.1 Terra significantly outperforms its predecessor and competes with GPT-6 Astra with medium reasoning effort, at a lower cost. With maximum reasoning effort, it consumes fewer tokens than Claude Opus 5.5. The model is also presented as making fewer factual errors and generating fewer hallucinations than previous versions.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

Positioning: intermediate model and choice for complex coding

OpenAI clarifies that for the most complex coding tasks, GPT-6.1 Sol and GPT-6 Astra remain preferable. GPT-6.1 Terra is described as "fast and efficient," combining detailed visual reasoning, precise computer usage, and clearer, more natural writing. This intermediate model succeeds GPT-5.6 Terra. GPT-6.1 Sol was introduced at DevDay 2026.

"Critical" level in cybersecurity and Astra safeguards

GPT-6.1 Terra reaches the "critical" risk level in cybersecurity according to OpenAI's internal grid, the highest level, and benefits from the safeguards designed for Astra. OpenAI also indicates that the model deviates less often from the scope of the assigned task and more honestly reports its limitations and potential bottlenecks.