GPT-5.6 Luna (Xhigh) vs Step 5 Preview
Which is better for the money: GPT-5.6 Luna (Xhigh) by OpenAI or Step 5 Preview by StepFun? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-25.
Trade-off
Verdict
Depends on your budget
Step 5 Preview scores 9.1 points higher (43.7 vs 34.6) but costs 6.6× as much to run the benchmark ($1190 vs $179; premium vs mid-range cost classes). They are not alternatives at one budget: pick Step 5 Preview when you need intelligence above 34.6, and GPT-5.6 Luna (Xhigh) when 34.6 is enough.
| Metric | GPT-5.6 Luna (Xhigh) | Step 5 Preview | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 34.6 | 43.7 | +26.3% Step 5 Preview is 26.3% more intelligent |
| Total AA cost (lower is better) | $179.33 | $1,190.26 | +563.7% Step 5 Preview is 6.6× more expensive |
| End-to-end time (lower is better) | 37.10s | 12.32s | −66.8% Step 5 Preview is 66.8% faster |
| Price per 1M tokens (in / out) | $0.20 / $1.20 | $1.00 / $2.70 | |
| Input modalities | Text, Image | Text, Image |
Step 5 Preview vs GPT-5.6 Luna (Xhigh): 26.3% more intelligence, 563.7% more cost to run the AA benchmark, and 66.8% faster end-to-end.
Today's Pareto frontier
Against the frontier
- GPT-5.6 Luna (Xhigh) is off the frontier: at the same or lower cost, Claude Haiku 5.5 (Xhigh) scores 41.20 for $158.
- Step 5 Preview is off the frontier: at the same or lower cost, GPT-6.1 Sol (Max) scores 51.80 for $1082.
GPT-5.6 Luna (Xhigh)Step 5 PreviewToday's Pareto frontierOpen-weights frontierEvery other model
Matchup
Compare something else
Agents:
Accept: text/markdown or ?format=md returns this page as Markdown. JSON: /api/v1/compare/gpt-5-6-luna-xhigh/step-5 · Markdown: /api/v1/compare/gpt-5-6-luna-xhigh/step-5?format=md · llms.txt · agent skill