Verdict
Built from the measurements below
- Accuracy is too close to call: GPT-5.3-Codex scored 82.5% and GPT-5.6 Terra 75.0% on the same 80 questions, a gap within what chance alone could produce.
- GPT-5.6 Terra is 1.6× cheaper per question as billed ($0.0063 vs $0.01), reasoning tokens included.
- Output speed is similar (95 vs 101 tokens per second).
- GPT-5.6 Terra starts answering sooner: 1.1 s to the first token, against 1.4 s for GPT-5.3-Codex.
- GPT-5.6 Terra accepts more context: 1.1M tokens against 400k.
Which should you choose?
Choose GPT-5.6 Terra for the lower cost per correct answer ($0.0084 against $0.013), a quicker first token and a 1.1M-token context window. The widest subject gap is in medicine: GPT-5.3-Codex answered 7 of 11 correctly, GPT-5.6 Terra 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 +7.5 points (z = 1.16). We call a winner only when z is at least 1.96, the usual 95% threshold. Speed, latency and price need a 10% margin.