DeepSeek V4 Flash 0731 (Max) vs Step 5 Preview
Which is better for the money: DeepSeek V4 Flash 0731 (Max) by DeepSeek or Step 5 Preview by StepFun? Intelligence, total benchmark cost, and end-to-end time from Artificial Analysis & other sources, snapshot 2026-10-24.
Trade-off
Verdict
Depends on your budget
Step 5 Preview scores 9.4 points higher (43.7 vs 34.3) but costs 6.0× as much to run the benchmark ($1190 vs $200; premium vs mid-range cost classes). They are not alternatives at one budget: pick Step 5 Preview when you need intelligence above 34.3, and DeepSeek V4 Flash 0731 (Max) when 34.3 is enough.
| Metric | DeepSeek V4 Flash 0731 (Max) | Step 5 Preview | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 34.3 | 43.7 | +27.4% Step 5 Preview is 27.4% more intelligent |
| Total AA cost (lower is better) | $199.63 | $1,190.26 | +496.2% Step 5 Preview is 6.0× more expensive |
| End-to-end time (lower is better) | 17.91s | 12.06s | −32.7% Step 5 Preview is 32.7% faster |
| Price per 1M tokens (in / out) | $0.14 / $0.28 | $1.00 / $2.70 | |
| Input modalities | Text | Text, Image |
Step 5 Preview vs DeepSeek V4 Flash 0731 (Max): 27.4% more intelligence, 496.2% more cost to run the AA benchmark, and 32.7% faster end-to-end.
Today's Pareto frontier
Against the frontier
- DeepSeek V4 Flash 0731 (Max) 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.
DeepSeek V4 Flash 0731 (Max)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/deepseek-v4-flash/step-5 · Markdown: /api/v1/compare/deepseek-v4-flash/step-5?format=md · llms.txt · agent skill