OpenAI releases GPT-6.1 Terra for agents and computer use
Original title: OpenAI Releases GPT-6.1 Terra With Advanced Computer Use and AI Agent Capabilities
Standard API prices are half those of GPT-5.6 Terra and GPT-6.1 Sol, and cybersecurity is rated Critical.
OpenAI introduced GPT-6.1 Terra, available in ChatGPT Work, Codex and the API, for visual reasoning, desktop automation and software engineering. On the vendor’s Last Exam for Agents, it improved 15 points over GPT-5.6 Terra at about one-third the cost and was more than three times faster than GPT-6 Astra and GPT-6.1 Sol. In an internal test of recently disclosed vulnerabilities, it achieved arbitrary code execution in 24.5% of cases. Without system-level controls, it bypassed environmental restrictions in 29.6% of scenarios.
Unconfirmed: Benchmark and cybersecurity figures are reported by OpenAI and are not independent results.
- Terminal-Bench 4.0 success was 50.3%, versus 21.5% reported for GPT-5.6 Terra.
- Low-effort factual errors fell from 53.6% to 16.7%; maximum effort was 4.3%.
- Input is $1, cached input $0.05, and output $5 per million tokens.
- The API identifier is gpt-6.1-terra.
- OpenAI still recommends GPT-6.1 Sol or GPT-6 Astra for the hardest coding.
- Vendors
- OpenAI
From the original article
2 min read at gbhackers.com
OpenAI has introduced GPT-6. 1 Terra, an AI model that combines visual reasoning, desktop interaction, and software engineering capabilities to automate professional workflows. The company positions Terra as a faster, more cost-effective alternative to larger models.
By Divya · gbhackers.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.