Data points: 135
Model compare
The readout for DeepSeek V4 Flash and Step 3.7 Flash, before the detailed comparison sheet.
Weighted outcome: Step 3.7 Flash. Benchmark capability categories carry 80%, while price, API performance, and availability carry 20%.
Decision read
Step 3.7 Flash
Step 3.7 Flash has the higher weighted result; Model A / B score 30 to 70.
Evidence depth
135 data points
Includes 32 benchmark rows, 1 audit samples, and 9 provider examples.
Selection signal
Start with Step 3.7 Flash
The charts below split 42 high-signal samples across speed, scores, and audit health.
Switch either side of this report to compare another model with the same LMSpeed data pipeline.
Select a different model to open a new comparison URL.
This report only uses LMSpeed data for DeepSeek V4 Flash and Step 3.7 Flash: pricing, speed aggregates, third-party benchmark scores, and shared provider samples.
| Model compare | DeepSeek V4 Flash | Step 3.7 Flash |
|---|---|---|
| Overall leader | Contender | Leading |
| Weighted overall score | 30.0 pts | 70.0 pts |
| Benchmark category leads | 1 categories | 4 categories |
| Operational advantages | Cheapest input price, First-token latency, Provider coverage | Average speed |
| Context window | 1.0M tokens | 256K tokens |
| Max output | 65.5K tokens | 256K tokens |
| Modalities | Input Text Output Text | Input Text |
The overall result weights benchmark capability categories at 80% and price, API speed/latency, and availability at 20%. Recent test volume does not affect the winner, and missing benchmark categories are excluded.
| Model compare | DeepSeek V4 Flash | Step 3.7 Flash |
|---|---|---|
| Developer | DeepSeek | StepFun |
| Released | Apr 2026 | May 2026 |
| Parameters | No data | No data |
| Tokenizer | DeepSeek | Other |
| Knowledge cutoff | No data | No data |
| OpenRouter ID | deepseek/deepseek-v4-flash | stepfun/step-3.7-flash |
| References | No data | No data |
This report only uses LMSpeed data for DeepSeek V4 Flash and Step 3.7 Flash: pricing, speed aggregates, third-party benchmark scores, and shared provider samples.
DeepSeek V4 Flash
DeepSeek V4 Flash is stronger in benchmark categories (Instruction following) and operational dimensions (Cheapest input price, First-token latency, Provider coverage).
Step 3.7 Flash
Step 3.7 Flash is stronger in benchmark categories (Agents, Coding, Reasoning, Knowledge) and operational dimensions (Average speed).
Third-party benchmark profile synced into LMSpeed; only metrics available for both models are shown.
Compare benchmark category scores on a 0-100 scale. Select a category to inspect the gap.
Avg. score
DeepSeek V4 Flash
45.8
Avg. score
Step 3.7 Flash
52
Step 3.7 Flash leads by 6.1
Step 3.7 Flash leads by 5.2
Step 3.7 Flash leads by 1.7
Step 3.7 Flash leads by 3.9
DeepSeek V4 Flash
No data
Step 3.7 Flash
DeepSeek V4 Flash leads by 2.3
Metric-level scores with benchmark source, rank depth, confidence, error, and evaluation date where available.
DeepSeek V4 Flash
winner58.0
Rank #39/83 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
53.0
Rank #51/83 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner$0.280/M
Rank #6/162 · confidence 4
Step 3.7 Flash
$1.15/M
Rank #42/162 · confidence 4
DeepSeek V4 Flash
winner$0.175/M
Rank #11/162 · confidence 4
Step 3.7 Flash
$0.438/M
Rank #34/162 · confidence 4
DeepSeek V4 Flash
winner$0.140/M
Rank #17/162 · confidence 4
Step 3.7 Flash
$0.200/M
Rank #27/162 · confidence 4
DeepSeek V4 Flash
winner7.1
Rank #29/73 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
2.3
Rank #45/73 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner86.7%
Rank #35/187 · confidence 4
Step 3.7 Flash
80.9%
Rank #73/187 · confidence 4
DeepSeek V4 Flash
winner27.8%
Rank #35/186 · confidence 4
Step 3.7 Flash
19.9%
Rank #57/186 · confidence 4
DeepSeek V4 Flash
63.0
Rank #53/89 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner63.7
Rank #50/89 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
56.9
Rank #22/26 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner59.5
Rank #20/26 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner56.2
Rank #22/58 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
39.6
Rank #40/58 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
52.6
Rank #31/37 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner56.3
Rank #22/37 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner44.9
Rank #41/96 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
40.0
Rank #58/96 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner51.8
Rank #44/64 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
41.5
Rank #56/64 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner42.0%
Rank #51/184 · confidence 4
Step 3.7 Flash
40.0%
Rank #68/184 · confidence 4
DeepSeek V4 Flash
winner95.8
Rank #1/90 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
84.4
Rank #28/90 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner40.3
Rank #26/99 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
30.3
Rank #58/99 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner89.4
Rank #27/96 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
80.9
Rank #61/96 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner32.1
Rank #27/96 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
19.9
Rank #53/96 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner37.2
Rank #33/90 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
25.4
Rank #54/90 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner1240.0
Rank #32/65 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
1214.0
Rank #40/65 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner79.2
Rank #5/86 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
67.3
Rank #46/86 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
95.0
Rank #19/84 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner98.5
Rank #2/84 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
45.1
Rank #5/5 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner47.2
Rank #4/5 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
57.8
Rank #17/28 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner67.1
Rank #6/28 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
47.8
Rank #12/21 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner49.5
Rank #10/21 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner31.1
Rank #15/54 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
21.5
Rank #33/54 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
73.2
Rank #16/24 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner75.8
Rank #15/24 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner54.4
Rank #18/50 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
51.6
Rank #20/50 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner1189.0
Rank #21/54 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
1017.0
Rank #33/54 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
winner34.4
Rank #22/50 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
25.9
Rank #33/50 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
56.9
Rank #31/45 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner59.5
Rank #26/45 · confidence 1 · eval date 2026-05-29
DeepSeek V4 Flash
46.6
Rank #36/52 · confidence 2 · eval date 2026-04-24
Step 3.7 Flash
winner51.7
Rank #32/52 · confidence 1 · eval date 2026-05-29
Latest completed audits from shared providers, with four safety and integrity score groups plus report links.
| Provider | DeepSeek V4 Flash | Step 3.7 Flash |
|---|---|---|
Winner: DeepSeek V4 Flash | DeepSeek V4 Flash deepseek-v4-flash Audit score 83 807280100 | Step 3.7 Flash step-3.7-flash No audit yet |
Speed aggregates and input/output pricing share each provider row for real API selection and migration cost checks.
| Provider | DeepSeek V4 Flash | Step 3.7 Flash |
|---|---|---|
60 tests | DeepSeek V4 Flash speed / latency 23 tok/s / 14357ms input / output No data | Step 3.7 Flash speed / latency 119 tok/s / 15574ms input / output No data |
20 tests | DeepSeek V4 Flash speed / latency 83 tok/s / 5114ms input / output No data | Step 3.7 Flash speed / latency N/A / N/A input / output No data |
10 tests | DeepSeek V4 Flash speed / latency 86 tok/s / 4077ms input / output No data | Step 3.7 Flash speed / latency N/A / N/A input / output No data |
10 tests | DeepSeek V4 Flash speed / latency 76 tok/s / 2331ms input / output No data | Step 3.7 Flash speed / latency N/A / N/A input / output No data |
10 tests | DeepSeek V4 Flash speed / latency 16 tok/s / 14964ms input / output No data | Step 3.7 Flash speed / latency N/A / N/A input / output No data |
DeepSeek V4 Flash deepseek-v4-flash speed / latency No data input / output $0.012/M/$0.023/M | Step 3.7 Flash step-3.7-flash speed / latency No data input / output $0/M | |
DeepSeek V4 Flash deepseek-v4-flash speed / latency No data input / output $0/M/$0/M | Step 3.7 Flash step-3.7-flash speed / latency No data input / output $0/M | |
DeepSeek V4 Flash deepseek-v4-flash speed / latency No data input / output $0/M/$0/M | Step 3.7 Flash step-3.7-flash speed / latency No data input / output $0.014/M | |
DeepSeek V4 Flash deepseek-v4-flash speed / latency No data input / output $0/request | Step 3.7 Flash step-3.7-flash-free speed / latency No data input / output $0/request |
Weighted outcome: Step 3.7 Flash. Benchmark capability categories carry 80%, while price, API performance, and availability carry 20%.
Output
| Features | Text inputText outputTool callingStructured outputsJSON modeReasoning | Text inputImage inputText outputTool callingStructured outputsJSON modeReasoning |
|---|
