DeepSeek: DeepSeek V4 Pro vs StepFun: Step 3.7 Flash
Based on weekly OpenRouter usage tracked by AI Tier List, DeepSeek: DeepSeek V4 Pro ranks #7 overall — ahead of StepFun: Step 3.7 Flash at #14. StepFun: Step 3.7 Flash is cheaper on input tokens at $0.20/1M vs $1.17/1M for DeepSeek: DeepSeek V4 Pro. DeepSeek: DeepSeek V4 Pro offers the larger context window at 1M tokens.
DeepSeek: DeepSeek V4 Pro vs StepFun: Step 3.7 Flash: which should you use?
Choose DeepSeek: DeepSeek V4 Pro if…
- It sees more real traffic — #7 overall on OpenRouter's weekly ranking, against #14 for StepFun: Step 3.7 Flash.
- Bigger context window — 1M vs 262K tokens.
- Higher Artificial Analysis intelligence index — 45.3 vs 30.9.
- Higher Artificial Analysis coding score — 59.4 vs 39.6.
- Higher Artificial Analysis agentic score — 37.8 vs 21.7.
Choose StepFun: Step 3.7 Flash if…
- Cheaper to read — $0.20 vs $1.17 per 1M input tokens, which is what dominates long-prompt and RAG workloads.
- Cheaper to write — $1.15 vs $2.34 per 1M output tokens, the line item that decides generation-heavy costs.
- It accepts image, video input, which DeepSeek: DeepSeek V4 Pro does not.
- It is the newer model — released May 2026 against Apr 2026.
Spec Comparison
| Spec | DeepSeek: DeepSeek V4 Pro | StepFun: Step 3.7 Flash |
|---|---|---|
| Author | deepseek | stepfun |
| Input price (per 1M tokens) | $1.17 | $0.20 |
| Output price (per 1M tokens) | $2.34 | $1.15 |
| Context window | 1M | 262K |
| Input modalities | text | text, image, video |
| Output modalities | text | text |
| AA Intelligence | 45.3 | 30.9 |
| AA Coding | 59.4 | 39.6 |
| AA Agentic | 37.8 | 21.7 |
| Released | Apr 2026 | May 2026 |
What DeepSeek: DeepSeek V4 Pro and StepFun: Step 3.7 Flash actually cost to run
On a read-heavy job — 1M tokens in, 100K out — DeepSeek: DeepSeek V4 Pro costs $1.40 and StepFun: Step 3.7 Flash costs $0.32.
Flip the shape — 100K in, 1M out — and it becomes $2.45 for DeepSeek: DeepSeek V4 Pro versus $1.17 for StepFun: Step 3.7 Flash.
Both figures are the published per-token rates multiplied out; caching and batch discounts are not included.
Usage Rank by Task
| Task | DeepSeek: DeepSeek V4 Pro | StepFun: Step 3.7 Flash |
|---|---|---|
| Overall | #7 | #14 |
| Memory Extraction | #8 | #5 |
| Multi-step Planning | #7 | — |
| Tool Dispatch | #8 | #8 |
| Web Search | #5 | — |
| Agent Workflows | #6 | #8 |
| Classification & Tagging | #10 | — |
| Debugging | #5 | #9 |
| DevOps Config | #8 | #9 |
| File Read/Write | #9 | #8 |
| Frontend UI | #9 | #8 |
| Code Implementation | #8 | #10 |
| Repo Analysis | #8 | #9 |
| Code Review & Security | #10 | #10 |
| Shell Execution | #1 | #8 |
About DeepSeek: DeepSeek V4 Pro and StepFun: Step 3.7 Flash
DeepSeek: DeepSeek V4 Pro
Built by deepseek · released Apr 2026 · 1M-token context
DeepSeek V4 Pro is a large-scale Mixture-of-Experts model from DeepSeek with 1.6T total parameters and 49B activated parameters, supporting a 1M-token context window. It is designed for advanced reasoning, coding,...
StepFun: Step 3.7 Flash
Built by stepfun · released May 2026 · 262K-token context
Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model. It pairs a 196B-parameter language backbone with a vision encoder for native image and video understanding, activating roughly 11B parameters...
Related comparisons
- DeepSeek: DeepSeek V4 Pro vs DeepSeek: DeepSeek V4 Flash 0731
- DeepSeek: DeepSeek V4 Pro vs Tencent: Hy3
- DeepSeek: DeepSeek V4 Pro vs OpenAI: GPT-5.6 Luna (batch)
- StepFun: Step 3.7 Flash vs DeepSeek: DeepSeek V4 Flash 0731
- StepFun: Step 3.7 Flash vs Tencent: Hy3
- StepFun: Step 3.7 Flash vs OpenAI: GPT-5.6 Luna (batch)
Get notified when DeepSeek: DeepSeek V4 Pro or StepFun: Step 3.7 Flash moves
Usage ranks and prices refresh weekly. We email you when they move — once a week, no spam.
FAQ
Which is better, DeepSeek: DeepSeek V4 Pro or StepFun: Step 3.7 Flash?
Based on weekly OpenRouter usage tracked by AI Tier List, DeepSeek: DeepSeek V4 Pro ranks #7 overall — ahead of StepFun: Step 3.7 Flash at #14. StepFun: Step 3.7 Flash is cheaper on input tokens at $0.20/1M vs $1.17/1M for DeepSeek: DeepSeek V4 Pro. DeepSeek: DeepSeek V4 Pro offers the larger context window at 1M tokens.
Which is cheaper, DeepSeek: DeepSeek V4 Pro or StepFun: Step 3.7 Flash?
StepFun: Step 3.7 Flash is cheaper at $0.20 per 1M input tokens. Output pricing: DeepSeek: DeepSeek V4 Pro $2.34, StepFun: Step 3.7 Flash $1.15.
Which is better for coding, DeepSeek: DeepSeek V4 Pro or StepFun: Step 3.7 Flash?
DeepSeek: DeepSeek V4 Pro. Artificial Analysis coding scores: DeepSeek: DeepSeek V4 Pro 59.4, StepFun: Step 3.7 Flash 39.6. Best coding-task usage rank this week: DeepSeek: DeepSeek V4 Pro #1, StepFun: Step 3.7 Flash #8.
Is DeepSeek: DeepSeek V4 Pro still ahead of StepFun: Step 3.7 Flash right now?
Yes. For the week of 2026-08-10, DeepSeek: DeepSeek V4 Pro ranks #7 overall on OpenRouter usage against #14 for StepFun: Step 3.7 Flash. Ranks refresh weekly, and the verdict on this page flips if they cross.
Usage ranks are weekly data aggregated by OpenRouter. Prices are USD per 1M tokens.