Compare/Step 3.5 Flash 2603 vs gpt-oss-20B (low)

Step 3.5 Flash 2603vsgpt-oss-20B (low)

Side-by-side comparison of pricing, 12 benchmarks, and generation speed.

StepFun

Step 3.5 Flash 2603

Input
$0/M
Output
$0/M
Speed
161 tok/s
TTFT
0.80s
OpenAI

gpt-oss-20B (low)

Input
$0.06/M
Output
$0.2/M
Speed
265 tok/s
TTFT
0.42s

Winner by Category

Cheaper
Step 3.5 Flash 2603
Faster (tok/s)
gpt-oss-20B (low)
Lower Latency
gpt-oss-20B (low)
Benchmarks (7-4)
Step 3.5 Flash 2603

Pricing Comparison

MetricStep 3.5 Flash 2603gpt-oss-20B (low)
Input ($/M tokens)$0$0.06
Output ($/M tokens)$0$0.2
Cost for 1M input + 100K output tokens:
Step 3.5 Flash 2603$0.00
gpt-oss-20B (low)$0.08

Speed Comparison

Output Speed (tokens/s) — higher is better
Step 3.5 Flash 2603
161 tok/s
gpt-oss-20B (low)
265 tok/s
Time to First Token (seconds) — lower is better
Step 3.5 Flash 2603
0.80s
gpt-oss-20B (low)
0.42s

Benchmark Comparison

Data from Artificial Analysis API — 12 benchmarks

Intelligence Index
38.520.8
Coding Index
34.614.4
Math Index
62.3
GPQA Diamond
82.6%61.1%
MMLU-Pro
71.8%
LiveCodeBench
65.2%
AIME 2025
62.3%
MATH-500
Humanity's Last Exam
22.6%5.1%
SciCode
38.5%34.0%
IFBench
66.5%57.8%
TerminalBench
32.6%4.5%
Step 3.5 Flash 26037 wins
4 winsgpt-oss-20B (low)

Frequently Asked Questions

Which is cheaper, Step 3.5 Flash 2603 or gpt-oss-20B (low)?

Step 3.5 Flash 2603 is cheaper overall. Its blended price (3:1 input/output ratio) is $0.00/M tokens vs $0.10/M for gpt-oss-20B (low).

Which model performs better on benchmarks?

Step 3.5 Flash 2603 wins 7 out of 12 benchmarks compared to 4 for gpt-oss-20B (low). See the detailed benchmark chart above for per-category results.

Which is faster for real-time applications?

gpt-oss-20B (low) generates tokens faster at 265 tok/s vs 161 tok/s. However, gpt-oss-20B (low) has lower time-to-first-token (0.42s vs 0.80s).

When should I use Step 3.5 Flash 2603 vs gpt-oss-20B (low)?

Choose based on your priorities: Step 3.5 Flash 2603 for lower cost, Step 3.5 Flash 2603 for stronger benchmark performance, and gpt-oss-20B (low) for faster generation. For latency-sensitive apps, check the TTFT comparison above.