AI BENCHY
Advertise here
#74

Step 3.7 Flash

Stepfun Release: 2026-05-29 Tested on: 2026-05-29 11:53 stepfun/step-3.7-flash::high
198B total (11B active)MoEOpen source
(high) (medium) (low)

Summary

Step 3.7 Flash scores 7.1 on AI BENCHY and ranks #74. It has 9.9 reliability, a 65.8% pass rate, $0.960 total cost, and 49.43s average response time.

Model facts

Researched on 2026-08-12

Reported
Parameters
198B total (11B active)
Architecture
MoE
Availability
Open source
License
Apache-2.0

Includes the vendor-reported vision component in the total.

Parameter source Availability source

Score

7.1

Consistency

8.2

Reliability

9.9

Total Cost (Current Price)

$0.960

Total Output Tokens

828,084

Total Input Tokens

0

Input Price

$0.200 / 1M

Output Price

$1.150 / 1M

Tests Correct

Wrong Tests: 9

Attempt pass rate: 65.8%

Flaky tests

4

Flaky tests had mixed outcomes across runs (at least one pass and one fail).

Response Time (avg)

49.43s

Response Time (max): 192.75s

Response Time (total): 988.58s

Wrong answer: 6 No answer: 3

Hamster playing table tennis

Prompt: Create a detailed SVG illustration of a hamster playing table tennis.

#74 Step 3.7 Flash

high
Cost
$0.007
Time
63.6s
Tokens
6,030 tok

Run history

Tested on Score Reliability Tests Correct Total Cost Compare
2026-08-14 02:45 Re-test 6.8 10.0 $1.229 Compare
2026-07-16 22:46 New test added 6.9 10.0 $1.207 Compare
2026-06-04 14:10 New test added 7.0 10.0 $1.148 Compare
2026-05-29 11:53 Initial run 7.1 9.9 $0.960 Current run

Charts

Choose the first model, then click a second model to open a side-by-side page.

Total Output Tokens

Score vs Total Output Tokens

Quick Compare

Step 3.7 FlashhighvsQwen3.6 FlashmediumStep 3.7 FlashhighvsSeed-2.0-CodemediumStep 3.7 FlashhighvsGemini 3.5 FlashnoneStep 3.7 FlashhighvsSolar Pro 4lowStep 3.7 FlashhighvsSeed-2.0-CodehighStep 3.7 FlashhighvsSolar Pro 4mediumStep 3.7 FlashhighvsQwen3.7 FlashmediumStep 3.7 FlashhighvsDeepSeek V4 PrononeStep 3.7 FlashhighvsKAT-Coder-Pro V2.5mediumStep 3.7 FlashhighvsGLM 5.3low

Category Breakdown

Category Score Consistency Tests Correct
Anti-AI Tricks 10.0 10.0
Coding 3.6 4.6
Combined 10.0 10.0
Data parsing and extraction 10.0 10.0
Domain specific 4.1 4.4
General Intelligence 5.5 10.0
Instructions following 9.8 10.0
Puzzle Solving 5.3 7.2
Tool Calling 10.0 10.0
Trivia 3.0 10.0

Compared models