Gemini 3.8 Flash High vs Step 5 Preview compared

Gemini 3.8 Flash High leads the intelligence index 30.8 to 28.5. Step 5 Preview costs 5% less on blended price. Benchmarks, sample bills, speed and a verdict.

Metric	Gemini 3.8 Flash (High)	Step 5 Preview
Intelligence Index	30.8	28.5
Input price (per 1M tokens)	$0.75	$1.00
Output price (per 1M tokens)	$3.75	$2.70
Output speed (tokens/s)	121	151
Time to first token	27s	0.91s
Release date	2026-09-02	2026-09-18

Metric

Gemini 3.8 Flash (High)

Step 5 Preview

Intelligence Index

30.8

28.5

Input price (per 1M tokens)

$0.75

$1.00

Output price (per 1M tokens)

$3.75

$2.70

Output speed (tokens/s)

121

151

Time to first token

27s

0.91s

Release date

2026-09-02

2026-09-18

On the headline score, Google's Gemini 3.8 Flash (High) and StepFun's Step 5 Preview are almost level. What separates them most is responsiveness; Gemini 3.8 Flash High takes 27 seconds to produce its first token, Step 5 Preview 0.91 seconds. Everything below is read from the benchmark, price and speed data in Franklin AI's catalogue; nothing here comes from a hands-on test.

How fast they respond

Gemini 3.8 Flash High generates about 121 tokens per second and Step 5 Preview about 151 tokens per second. Step 5 Preview is ahead by 25% on throughput. Before any text appears, Gemini 3.8 Flash High takes 27 seconds and Step 5 Preview 0.91 seconds. Gemini 3.8 Flash High waits 30 times as long before its first token.

Add the two numbers up for a 1,500-token reply and Step 5 Preview finishes first: 40 seconds for Gemini 3.8 Flash High, 11 seconds for Step 5 Preview.

What they cost

Per million tokens, Gemini 3.8 Flash High charges $0.75 for input and $3.75 for output, while Step 5 Preview charges $1.00 and $2.70. The rates cross over. Gemini 3.8 Flash High is cheaper to feed and Step 5 Preview is cheaper to hear back from.

For a sample workload of 1 million input tokens and 200,000 output tokens, Gemini 3.8 Flash High would cost $1.50 and Step 5 Preview $1.54, a difference of $0.040. A bigger example, a month of coding-assistant traffic at 30 million input and 3 million output tokens, costs $33.75 on Gemini 3.8 Flash High and $38.10 on Step 5 Preview: a gap of $4.35. On a 3:1 blend of input to output tokens, that works out to $1.50 versus $1.425 per million.

Benchmark scores

The catalogue puts Gemini 3.8 Flash High at 30.8 and Step 5 Preview at 28.5 on its intelligence index. It is a small margin for an index that averages many evaluations. In relative terms, Gemini 3.8 Flash High scores 8% higher.

Among 109 active scored models, Gemini 3.8 Flash High sits 50th and Step 5 Preview 61st.

The catalogue does not yet split either model's score into coding and math indexes.

Which is newer

Gemini 3.8 Flash High came first, on September 2, 2026; Step 5 Preview followed on September 18, 2026. That is close enough to count them as the same generation.

Score per dollar

Per point of intelligence index, Gemini 3.8 Flash High costs $0.049 of blended price per million tokens and Step 5 Preview $0.050. On that measure they are priced in proportion to their scores.

FAQ

Is Gemini 3.8 Flash High better than Step 5 Preview?

Neither wins everywhere: Gemini 3.8 Flash High scores higher on the intelligence index, 30.8 to 28.5; Step 5 Preview is 5% cheaper on blended price; and Step 5 Preview streams output 1.3 times as fast.

Are Gemini 3.8 Flash High and Step 5 Preview from the same company?

Gemini 3.8 Flash High is made by Google and Step 5 Preview by StepFun. Gemini 3.8 Flash High was released on September 2, 2026 and Step 5 Preview on September 18, 2026.

Which model has lower latency?

Step 5 Preview generates output faster, at 151 tokens per second against 121 tokens per second. Step 5 Preview reaches its first token sooner (0.91 seconds against 27 seconds).

Is Step 5 Preview cheaper than Gemini 3.8 Flash High?

Step 5 Preview is cheaper on a blended basis. Gemini 3.8 Flash High lists $0.75 per million input tokens and $3.75 per million output tokens; Step 5 Preview lists $1.00 and $2.70. For 1 million input and 200,000 output tokens that is $1.50 against $1.54.

Which is newer, Gemini 3.8 Flash High or Step 5 Preview?

Step 5 Preview is newer. It was released on September 18, 2026, 16 days after Gemini 3.8 Flash High (September 2, 2026).

Final whistle

Short answer: Step 5 Preview is the stronger all-round pick on this data, while Gemini 3.8 Flash High makes sense in narrower cases. Pick Step 5 Preview if users are waiting on the first word and long outputs need to finish quickly. Pick Gemini 3.8 Flash High if you need the higher overall score (30.8 against 28.5).

Where the numbers come from

The scores, speed and prices in this comparison come from independent third-party model evaluations that Franklin AI syncs each day, and the article was written from a snapshot of them. For the latest results, open the Gemini 3.8 Flash (High) and Step 5 Preview pages. The AI Benchmark methodology explains how that data is collected and shown.