Mistral Small 4 (Reasoning) vs Step 5 Preview
Which is better for the money: Mistral Small 4 (Reasoning) by Mistral 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 32.4 points higher (43.7 vs 11.3) but costs 28× as much to run the benchmark ($1190 vs $42.5; premium vs budget cost classes). They are not alternatives at one budget: pick Step 5 Preview when you need intelligence above 11.3, and Mistral Small 4 (Reasoning) when 11.3 is enough.
| Metric | Mistral Small 4 (Reasoning) | Step 5 Preview | Difference |
|---|---|---|---|
| Intelligence (higher is better) | 11.3 | 43.7 | +286.7% Step 5 Preview is 3.9× more intelligent |
| Total AA cost (lower is better) | $42.50 | $1,190.26 | +2700.6% Step 5 Preview is 28× more expensive |
| End-to-end time (lower is better) | 14.50s | 12.06s | −16.8% Step 5 Preview is 16.8% faster |
| Price per 1M tokens (in / out) | $0.15 / $0.60 | $1.00 / $2.70 | |
| Input modalities | Text, Image | Text, Image |
Step 5 Preview vs Mistral Small 4 (Reasoning): 286.7% more intelligence, 2700.6% more cost to run the AA benchmark, and 16.8% faster end-to-end.
Today's Pareto frontier
Against the frontier
- Mistral Small 4 (Reasoning) is off the frontier: at the same or lower cost, GPT-6 Luna (Medium) scores 29.90 for $31.2.
- Step 5 Preview is off the frontier: at the same or lower cost, GPT-6.1 Sol (Max) scores 51.80 for $1082.
Mistral Small 4 (Reasoning)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/mistral-small-4/step-5 · Markdown: /api/v1/compare/mistral-small-4/step-5?format=md · llms.txt · agent skill