OpenAI rolls out budget-priced mid-tier GPT-6.1 Terra

What's happening

OpenAI has launched GPT-6.1 Terra, presented as a budget-priced mid-tier variant that in some tasks rivals pricier models while reportedly costing about a tenth of Astra's API price. Public benchmarks and community tests show mixed results: Terra scored 48 on an Artificial Analysis benchmark, matched or produced similar writing scores to many GPT-5.6 Terra settings while lowering cost in a YouTube script study, and struggled on hard algorithmic contest problems where smaller models also underperformed. Community comparisons note GPT-6.1 Sol often outperforms Terra on many benchmarks at nearly the same cost, and Sol achieved 72% as an agent in a Salesforce task suite (behind other models).

Why it's trending

Multiple independent benchmarks and community tests just surfaced after the launch, highlighting cost versus performance tradeoffs and sparking debate over Terra versus Sol.

SignalHolding at its usual pace, confirmed across 3 independent source types.