OpenAI Launches GPT-6.1 Terra: Cheaper, Smarter, Faster
OpenAI introduced GPT-6.1 Terra, a model balancing visual reasoning, computer use, and natural language generation at reduced cost. Standard API pricing is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol, with rates of $1 per million input tokens and $5 per million output tokens. On professional task benchmarks, Terra scores 15 percentage points higher than its predecessor while costing one-third as much, and matches GPT-6 Astra’s performance at one-tenth the per-token price.
The model demonstrates substantial improvements in factuality and coding. Internal error-inducing prompt benchmarks show error rates dropping from 53.6% to 16.7% at low reasoning effort—a reduction of nearly 37 percentage points. At maximum reasoning effort, errors fall to 4.3%. For terminal command-line coding problems, GPT-6.1 Terra solves 50.3% of tasks at max reasoning, compared to GPT-5.6 Terra’s published score of 21.5%.
The model executes tasks over three times faster than both GPT-6 Astra and GPT-6.1 Sol. GPT-6.1 Terra meets the Critical threshold in cybersecurity safeguards, shows improved alignment compared to its predecessor, and demonstrates strong resistance to prompt injection attacks. Now available to ChatGPT Work, Codex, and API users, Terra positions efficient capability as viable for production-scale workloads.
Read the full article
Continue reading at Hacker News →This is an AI-generated summary. Read the original for the full story.