Verdict
Built from the measurements below
- Accuracy is too close to call: GPT-5.4 scored 72.5% and GPT-6.1 Terra 73.8% on the same 80 questions, a gap within what chance alone could produce.
- GPT-6.1 Terra is 8.9× cheaper per question as billed ($0.0013 vs $0.011), reasoning tokens included.
- GPT-6.1 Terra generates text 2.2× faster: 122 tokens per second against 55.
- GPT-5.4 starts answering sooner: 1.7 s to the first token, against 2.4 s for GPT-6.1 Terra.
Which should you choose?
Choose GPT-5.4 for a quicker first token. Choose GPT-6.1 Terra for the lower cost per correct answer ($0.0017 against $0.016), faster streaming and shorter outputs (about 195 tokens per question against 706). The widest subject gap is in medicine: GPT-6.1 Terra answered 6 of 11 correctly, GPT-5.4 4 of 11. With so few questions per subject, treat this as a lead to test on your own prompts, not a result.
Accuracy gap -1.3 points (z = -0.18). We call a winner only when z is at least 1.96, the usual 95% threshold. Speed, latency and price need a 10% margin.