StepFun API pricing: Step 5 Preview, 3.7 Flash, rate limits

StepFun API pricing explained

Updated 2026-11-01. Official StepFun prices, LMArena ranks.

Quick answer

Step 5 Preview, StepFun's newest model, costs $1 input, $0.10 cached and $2.70 output per 1M tokens. Step 3.7 Flash costs $0.20 / $1.15 and Step 3.5 Flash is the cheapest at $0.10 / $0.30, per StepFun's official pricing page. There's no long-context price step, and your rate limits go up with how much you've topped up, from 100 requests a minute under $15 to 2,600 at $1,500 or more.

Model	Type	Input	Cached	Output	Cache saving
Step 5 Preview	Multimodal reasoning, preview	$1	$0.10	$2.70	90%
Step 3.7 Flash	Multimodal reasoning	$0.20	$0.04	$1.15	80%
Step 3.5 Flash	Reasoning (text)	$0.10	$0.02	$0.30	80%

USD per 1M tokens, StepFun Open Platform pay-as-you-go list price, checked 2026-11-01. Cached means a cache hit; for Step 5 Preview the input price already includes writing new content to the cache.

Which StepFun model should you use?

Step 5 Preview for the best quality StepFun sells. Step 5 Preview ranks #32 of 51 priced models on Text Arena and #28 of 50 on Code Arena, at $1 in and $2.70 out. It's a preview, so expect changes before a stable release.
Step 3.7 Flash for cheap multimodal work. Step 3.7 Flash takes images like Step 5 Preview at $0.20 / $1.15. LMArena doesn't rank it yet.
Step 3.5 Flash for bulk text. Step 3.5 Flash is $0.10 / $0.30, number 2 of 77 on our full price list, but it ranks #51 of 51 on Text Arena, so test it on your own prompts first. StepFun also lists a dated build, step-3.5-flash-2603, at the same price.

How StepFun prompt caching is priced

StepFun lists one cache-hit price per model, which works out to 90% off input on Step 5 Preview, 80% on Step 3.7 Flash and 80% on Step 3.5 Flash. On Step 5 Preview there is no separate cache-write fee, because the normal input price already covers it, per StepFun's pricing page. Output tokens include the model's reasoning as well as the final answer, so long thinking runs cost more than the visible reply suggests.

StepFun API rate limits by tier

Pay-as-you-go limits depend on how much cash you've topped up in total. Vouchers don't count, and individual accounts must verify their identity first. StepFun says it may lower limits for a while when capacity runs short.

Tier	Total top-ups	Concurrency	RPM	TPM
V0	Under $15	5	100	500,000
V1	$15 to $69	20	400	2,000,000
V2	$70 to $299	30	600	3,000,000
V3	$300 to $1,499	40	800	4,000,000
V4	$1,500 and above	130	2,600	13,000,000

Individual accounts, from StepFun's official pricing page. Enterprise tiers go through sales. Step Plan subscriptions use a monthly credit allowance instead.

Search, storage and other add-ons

Tool calls cost extra on top of tokens: internet search is $0.006 per call, image search $0.02 per call and file storage $0.08 per GB per day, per StepFun. Images you send count as input tokens, 169 tokens per image by default.

StepFun image models were retired

StepFun retired its image generation and editing models, step-2x-large and step-image-edit-2, on October 10, 2026, and those APIs no longer serve requests. If you still call them, you need another provider for images. The text, vision and speech models are still on sale.

What a month costs

List price, no discounts, our math. Medium is 20M input and 5M output tokens a month; heavy is 200M and 50M.

Model	Medium month	Heavy month
Step 3.5 Flash	$3.50	$35.00
Step 3.7 Flash	$9.75	$97.50
Step 5 Preview	$33.50	$335

StepFun against the models people compare it with

Step 5 Preview vs DeepSeek V4 Pro: $1.43 vs $1.98 blended.
Step 3.5 Flash vs DeepSeek V4.1 Flash: $0.15 vs $0.52 blended.
All Chinese labs side by side: best Chinese LLM APIs.

Sources

FAQ

How much does the StepFun API cost?

Step 5 Preview costs $1 per 1M input tokens, $0.10 cached and $2.70 output. Step 3.5 Flash is the cheapest at $0.10 in and $0.30 out, as of 2026-11-01.

What is the cheapest StepFun model?

Step 3.5 Flash at $0.10 input and $0.30 output per 1M tokens, with cache hits at $0.02.

What are StepFun's API rate limits?

They depend on your total top-ups. Under $15 you get 5 concurrent requests, 100 RPM and 500,000 TPM; at $1,500 or more you get 130 concurrent, 2,600 RPM and 13M TPM.

Does StepFun charge for reasoning tokens?

Yes. StepFun counts the model's reasoning and the final answer as output tokens, both billed at the output price.

Can I still generate images with StepFun?

No. StepFun retired step-2x-large and step-image-edit-2 on October 10, 2026, and its image generation and editing APIs stopped serving requests.

Is Step 5 Preview better than Step 3.5 Flash?

On LMArena, yes by a wide margin: Step 5 Preview ranks #32 of 51 on Text Arena and Step 3.5 Flash #51 of 51. Flash costs 9x less on output, so it suits simple bulk jobs.