OpenAI launches GPT-6.1 Terra model with reduced pricing and enhanced capabilities for computer agents
OpenAI has officially launched GPT-6.1 Terra, a new artificial intelligence model designed to enhance computer agent capabilities. The company announced that the model offers significantly lower pricing and improved performance compared to previous iterations and some competitors, aiming to make advanced AI more accessible for everyday computer work and enterprise automation. This release follows OpenAI's pattern of introducing more cost-effective models. Many outlets report that GPT-6.1 Terra achieves performance levels comparable to the more powerful GPT-6 Astra model, but at a fraction of the cost. However, some independent analyses suggest that while fast and cheap, it may still trail leading AI models in specific tasks, such as accounting benchmarks. Concerns regarding safety evaluations were also raised, with reports indicating that while generally safer, the model showed more destructive-action flags in tests. The central claims regarding performance and cost are primarily based on OpenAI's own announcements and benchmarks.
82 articles from 74 outlets covered this story. Their coverage differs on 4 points. The underlying claim is sourced from a shipped.
What do all outlets agree on?
74 outlets covered “OpenAI launches GPT-6.1 Terra model with reduced pricing and enhanced capabilities for…”. All of them report the following:
- OpenAI launched GPT-6.1 Terra
- It is a new AI model
- It is designed for computer use and AI agents
- It offers lower API pricing
- It provides improved speed and performance
Did outlets disagree about this?
Yes. Coverage of “OpenAI launches GPT-6.1 Terra model with reduced pricing and enhanced capabilities for…” differs on 4 points. Each account below is how a different outlet described the same event:
GPT-6.1 Terra achieves performance comparable to GPT-6 Astra.
Digital Trends, Interesting Engineering, Yahoo Tech, Tech Edition (claim Astra-like results); Yahoo Finance UK, The National Law Review, XenoSpectrum, Decrypt, Cryptonews.net (report it trails leading models or is 'half-good')
GPT-6.1 Terra is three times faster than Astra.
Mezha (claims); other outlets do not corroborate this specific speed comparison to Astra, focusing more on general speed improvements or efficiency over Claude Opus.
GPT-6.1 Terra's safety profile.
Help Net Security (claims it meets critical cyber threshold); The New Stack, BleepingComputer, Mezha.net, SC Media (report it shows more destructive-action flags in tests despite being safer overall).
GPT-6.1 Terra's position among top AI models.
Marketing4eCommerce (claims it 'enters the top 10 most intelligent AI models'); Yahoo Finance UK, The National Law Review, XenoSpectrum, Yahoo Tech, Decrypt, Cryptonews.net (state it 'trails leading AI models' or is 'half-good').
Which figures do outlets report differently?
3 figures in this story are reported with conflicting values:
Cost Reduction (general)50% lower pricing vs half the price
Cost Reduction (vs GPT-6.1 Sol)half the rate of a model OpenAI shipped 27 days ago vs halving GPT-6.1 Sol’s API price
Benchmark Scores48 vs 52 vs 53
Which outlets covered this?
All 82 articles found on this story, grouped by the stance of the piece. Every link goes to the original publisher.
optimist16
pragmatist62
safety4