Foundation Models
OpenAI Launches GPT-6.1 Terra With 50% Price Cut and 50.3% Terminal-Bench
Priced at $1 per million input tokens and $5 per million output tokens, the small reasoning model cuts max-effort error rates to 4.3% and crosses OpenAI’s Critical cybersecurity threshold.
Your listening Queue & catch up
Add a story or build a catch-up queue to start listening.
OpenAI has launched GPT-6.1 Terra, a new compact reasoning model that cuts API pricing by 50% compared with both its predecessor GPT-5.6 Terra and the larger GPT-6.1 Sol while delivering substantial gains on agentic coding and terminal execution benchmarks.
In the OpenAI API, GPT-6.1 Terra is priced at $1 per 1 million input tokens, $0.05 per 1 million cached input tokens, and $5 per 1 million output tokens. OpenAI is rolling out the model across ChatGPT and Codex alongside the API release.
Terminal-Bench 4.0 Gains and Critical Cybersecurity Threshold
On Terminal-Bench 4.0, GPT-6.1 Terra scored 50.3% at maximum reasoning effort, more than doubling the 21.5% score recorded by GPT-5.6 Terra. Internal reliability evaluations also showed sharp reductions in hallucination and task failure rates: compared with GPT-5.6 Terra's 53.6% baseline error rate, GPT-6.1 Terra reduced errors to 16.7% at low reasoning effort and 4.3% at maximum effort.
OpenAI disclosed that GPT-6.1 Terra is its first small-class model to reach the Critical capability threshold for cybersecurity under the company's Preparedness Framework, triggering mandatory defensive deployment controls and access monitoring.
Sources
Comments (0)
No comments yet. Be the first to share a thought on this story.