GPT-6.1 Terra: API Down 50%, Expanded Quotas and Astra Security

Key Takeaways

1OpenAI launches GPT-6.1 Terra with API pricing 50% lower than GPT-5.6 Terra and GPT-6.1 Sol

2The model covers more tasks for the same quota in Work and Codex

3OpenAI claims advancements in vision, agents, and reduction of factual errors

4Access to GPT-6.1 Terra on ChatGPT Work or Codex requires a Plus, Pro, Enterprise, or Edu subscription

Why it matters : GPT-6.1 Terra aims to make AI automation more accessible and reliable for professionals while keeping costs under control.

OpenAI launches GPT-6.1 Terra with API pricing announced to be 50% lower than neighboring models and quotas that extend further into Work and Codex. The company highlights advancements in vision, agent usage, a reduction in errors, and cyber protections aligned with GPT-6 Astra.

OpenAI aligns Terra with the protection level of GPT-6 Astra

OpenAI specifies that GPT-6.1 Terra has reached a risk threshold requiring the application of specific cybersecurity measures. The model is therefore protected by the same safeguards as GPT-6 Astra. According to an internal analysis conducted by OpenAI, the error rate of the model, when operating at full capacity, would be comparable to those observed for GPT-6 Astra and GPT-6.1 Sol.

Usage cost: API rates -50% and extended quotas in Work and Codex

On OpenAI's API, GPT-6.1 Terra is offered at a price 50% lower than that of GPT-5.6 Terra and GPT-6.1 Sol. OpenAI notes that in Work and Codex, the same usage quota allows for more tasks than with GPT-5.6 Terra, and significantly more than with GPT-6.1 Sol or GPT-6 Astra. The model is presented as enabling users to accomplish more tasks with the same subscription. To access GPT-6.1 Terra on ChatGPT Work or Codex, users must have a Plus, Pro, Enterprise, or Edu subscription. On ChatGPT Plus, it is expected that users will be able to do more without exhausting their quota. This approach addresses the issue of rising costs associated with high token consumption, which explains the interest in optimized intermediate models.

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

Claimed capabilities: vision, agents, and reduction in factual errors

OpenAI emphasizes the capabilities of GPT-6.1 Terra in visual understanding and for agent tasks. The company claims to have reduced factual errors compared to the previous version and mentions a significant decrease in errors. The evolution from GPT-5.6 Terra is presented as substantial. Furthermore, according to a ranking by Artificial Analysis based on multiple evaluations, GPT-6.1 Terra is approaching the level of GPT-6.1 Sol.

Context: competing models and pressure on cost-effectiveness

Anthropic has developed Claude Sonnet 5.5, whose performance is close to Opus 5.5 while being more economical. On its part, OpenAI offers GPT-6.1 Terra, an intermediate model providing good performance at a reduced cost. The cost-effectiveness ratio is also a major issue for the competitiveness of AI solutions, especially since some Chinese open models are considered financially advantageous.