Interactive image editing
Products that need a fast preview after a bounded natural-language edit.
Independent tool overview
Step Image Edit 2 at a glance
Step Image Edit 2 is StepFun's lightweight API model for both text-to-image generation and instruction-based image editing. Its standout proposition is a $0.003 per-image price and a vendor-reported 1–2 second editing response, but the broad content license in StepFun's terms makes data governance a serious procurement consideration.
Visit the official Step Image Edit 2 site ↗
- Developer
- StepFun Technology
- Type
- Hosted API for text-to-image generation and instruction-based image editing
- Model size
- 3.5B architecture in StepFun marketing; documented as under 6B
- Vendor latency
- 1–2 seconds per editing task; production end-to-end latency can differ
- Prompt limit
- 512 characters
- Input limit
- Up to 4096 × 4096 for image editing
- Edit output
- Returned at the same dimensions as the input image
- Price
- $0.003 per generated or edited image
- Data warning
- Current terms grant StepFun a broad perpetual license to prompts, outputs, and API logs
- Last reviewed
- August 31, 2026
Overview
What Step Image Edit 2 is
Step Image Edit 2 exposes one model through separate generation and editing endpoints. Developers can create a new image from text or upload an image and describe a change such as a pose adjustment, clothing replacement, object edit, or text addition.
StepFun presents the model as a 3.5-billion-parameter lightweight architecture in its current platform marketing and places it in the under-6B class in the model documentation. The documentation claims 1–2 second editing, a 512-character prompt limit, input images up to 4096 by 4096, and a price of $0.003 for either a generation or edit.
The model offers a useful combination of speed, low price, text-rendering mode, deterministic seeds, and same-size edited output. These qualities make it attractive for interactive prototypes, high-volume variant generation, and bounded product-photo or creative operations where every result is reviewed.
The procurement tradeoff is unusually important. StepFun's August 7, 2026 terms say customers own outputs to the extent permitted by law, but also grant StepFun a perpetual, irrevocable, transferable, sublicensable license to use prompts, outputs, and associated API logs broadly, including exploitation and commercialization. Early-access and startup programs can additionally use this data for model training. Do not send sensitive or proprietary images without a contract and data-processing arrangement that matches the intended use.
Use cases
Who Step Image Edit 2 is best for
The strongest fit depends on the job you need the product to complete, not the size of its feature list.
High-volume visual variants
Teams generating many low-cost candidates for a human to compare and approve.
Product and campaign mockups
Rapidly test outfits, backgrounds, object placement, signs, labels, and composition changes with non-sensitive assets.
Text-in-image experiments
Use the optional text mode when a design includes a short sign, caption, or label, then verify every character.
Latency-sensitive prototypes
Evaluate an under-6B hosted model where response speed and per-image cost matter more than maximum fidelity.
Capabilities
Core Step Image Edit 2 features
Unified model
One model supports both new image generation and edits to an uploaded image.
Instruction editing
Changes pose, clothing, objects, setting, composition, or visible text from a natural-language prompt.
Fast response
StepFun reports 1–2 second editing for its optimized hosted model.
Low fixed image price
Both generation and editing are listed at $0.003 per image.
Text-rendering mode
An optional text_mode parameter is intended to improve scenarios that require visible text.
Seed and sampling controls
API parameters include seed, steps, guidance scale, response format, and selected generation sizes.
Same-size editing
Edited results preserve the input image dimensions, up to the documented 4096 by 4096 limit.
Process
How the Step Image Edit 2 workflow works
Step 1
Classify the asset
Before uploading, decide whether the image contains confidential designs, unreleased products, personal data, biometric identifiers, minors, client material, or third-party rights.
Step 2
Confirm contract and consent
Use only assets the organization can legally process and obtain explicit consent for identifiable people; resolve StepFun's content license and program-specific training terms.
Step 3
Define a narrow edit
State what may change, what must remain fixed, required text, output dimensions, brand constraints, and prohibited additions within the 512-character limit.
Step 4
Generate a small batch
Record the model, prompt, seed, settings, input hash, output, latency, and cost for reproducibility.
Step 5
Compare against the source
Inspect identity, geometry, hands, objects, logos, text, color, lighting, unintended removals, and changes outside the requested region.
Step 6
Run rights and safety review
Check consent, misleading edits, trademarks, copyrights, publicity rights, prohibited impersonation, and whether AI disclosure is required.
Step 7
Approve before delivery
Keep a human approval gate before publishing, selling, sending to a client, updating a product catalog, or making a high-stakes decision.
Cost
Step Image Edit 2 pricing and free plan
StepFun lists the same fixed price for text-to-image generation and image editing. The platform is prepaid, can provide temporary trial credits, and applies request limits based on cumulative account top-ups.
Text-to-image
$0.003/image
Create one image from a text prompt through the generation endpoint.
- Fixed listed price per image
- Supported output sizes are predefined in the API documentation
- Storage, review, retries, and application infrastructure are separate
Image editing
$0.003/image
Edit one uploaded image through the image-edits endpoint.
- Output preserves input dimensions
- Input can be up to 4096 × 4096
- Each candidate and retry is a separate output to budget
Platform access
Prepaid
Accounts add balance before use; temporary trial credits may be offered under current platform rules.
- Base tier starts at five concurrent requests and ten requests per minute
- Higher cumulative top-ups increase published rate limits
- Pricing and limits can change with platform notice
- Contact StepFun for higher or more stable capacity
Pricing checked . Check current pricing at the source ↗
Assessment
Step Image Edit 2 strengths and limitations
Where it stands out
- Very low $0.003 per-image API price
- Generation and editing are available through one model
- Vendor-reported 1–2 second editing is suitable for responsive prototypes
- Supports large editing inputs up to 4096 by 4096
- Same-size edited output simplifies some application pipelines
- Optional text mode addresses a common weakness in image generation
- Seed and sampling parameters make controlled comparison easier
- Published rate-limit tiers help teams estimate scaling requirements
What to consider
- Benchmark leadership and comparisons with larger models are vendor claims; the reviewed documentation did not provide a reproducible benchmark table and evaluation protocol
- A 1–2 second model response does not include every production factor such as upload, queueing, network, moderation, serialization, and application latency
- The 512-character prompt limit constrains long, highly detailed edit specifications
- Same-size output is not the same as preserving every subject, pixel, logo, product detail, or identity
- Visible text can still be misspelled, distorted, or semantically wrong even with text mode enabled
- Fast low-cost output can contain anatomy, geometry, lighting, reflection, object-count, or consistency errors
- StepFun's terms grant a broad perpetual, irrevocable, transferable, sublicensable license to use prompts, outputs, and API logs, including commercialization
- Startup and other early-access programs may use prompts, outputs, and logs for training and fine-tuning; StepFun advises users who object not to join or to seek an alternative arrangement
- The current privacy policy says API inputs can include text, images, and voice and advises against sensitive, confidential, proprietary, or unauthorized information
- IP addresses are listed with a three-year retention period in the current privacy policy
- Users must obtain consent before submitting another person's photo for biometric processing or other edits and must not impersonate without legal right
- Output ownership “as between” the user and StepFun does not guarantee copyright protection, uniqueness, noninfringement, or clearance
- The terms prohibit representing AI output as human-generated and restrict high-stakes automated decisions
- Hosted-only access creates vendor, jurisdiction, policy, pricing, and availability dependency
Compare
Step Image Edit 2 alternatives
The right alternative depends on the specific output, workflow, controls and budget your project requires.
Design
FLUX.2 [klein]
Choose FLUX.2 Klein when you want another ultra-fast image-generation and editing model to benchmark on latency, quality, and deployment fit.
Explore FLUX.2 [klein] →Design
Flux 1 Kontext
Choose FLUX.1 Kontext dev when open weights and local control matter more than StepFun's very low hosted-API price.
Explore Flux 1 Kontext →Content Creator
Nano Banana 2 Lite
Choose Nano Banana 2 Lite when a Google-hosted, high-volume image model is a better fit for the surrounding platform and governance requirements.
Explore Nano Banana 2 Lite →Questions
Step Image Edit 2 FAQs
What is Step Image Edit 2?
It is StepFun's lightweight hosted model for both text-to-image generation and natural-language editing of an uploaded image.
How much does Step Image Edit 2 cost?
StepFun listed both text-to-image generation and image editing at $0.003 per image on August 31, 2026.
How fast is it?
StepFun claims 1–2 second editing. Measure complete application latency yourself because uploads, queues, network, moderation, and response processing add time.
What image sizes does it support?
Editing inputs can be up to 4096 by 4096 and edits preserve the input dimensions. Text-to-image generation supports the predefined sizes listed in StepFun's API documentation.
Can it add readable text?
The API includes an optional text_mode intended for text-rendering scenarios. Every character, number, symbol, and brand mark still needs visual review.
Does StepFun train on my images?
The current terms grant StepFun broad rights to prompts, outputs, and API logs. They specifically say startup and other early-access programs can use Usage Data for training and fine-tuning. Obtain contract-specific terms before sending confidential or proprietary assets.
Who owns the output?
StepFun's terms say the customer owns output as between the customer and StepFun to the extent permitted by law, while also granting StepFun a broad perpetual license to use the content. That does not guarantee copyright protection, uniqueness, or clearance.
Can I edit a person's photo?
Only with the necessary rights, consent, and lawful purpose. StepFun's terms prohibit submitting another person's image without consent for biometric processing or other purposes and prohibit unauthorized impersonation.
Are the benchmark claims independently proven?
Treat them as vendor claims. The reviewed model page asserts leadership within its parameter tier but did not include the full reproducible benchmark dataset, scoring method, and independent evaluation needed to generalize the result.
Bottom line
Our Step Image Edit 2 verdict
Step Image Edit 2 is compelling for teams that need extremely low-cost, fast generation and editing through a simple API. The model should earn a benchmark slot, not automatic adoption. Its broad contractual license to prompts and outputs is the decisive concern: use non-sensitive test assets first, negotiate suitable data terms, and require human visual and rights review before production.
Visit Step Image Edit 2 website ↗Similar Design Tools
View all Design AI tools →
Adobe Firefly AI Assistant
Adobe's conversational creative agent for planning, creating, and editing content across Firefly and Creative Cloud apps.
Recraft V4.1
Recraft V4.1 - Upgraded image model with enhanced photorealism, illustration, and prompt adherence
Ideogram Custom Models
Ideogram Custom Models -Fine-tuning tool for training on-brand image models from your own assets
Ideogram 4.0
Ideogram 4.0 - Ideogram's new top-rated open-weight image model
AI training for the future of work.
Get access to all our AI courses, hundreds of real-world AI use cases, live expert-led workshops, an exclusive network of AI early adopters, and more.
AI Courses
Get unlimited access to all of our current & upcoming industry-specific AI courses for the duration of your subscription.
Daily Guides
To keep up with the rapid pace of AI, our team publishes AI implementation guides daily. Our library contains 300+ practical use cases to automate real-world work.
Workshops
Join weekly, live, interactive sessions with industry leaders who are at the forefront of AI for hands-on implementation guidance and exclusive insights.
Community
Network with an exclusive community of AI-first professionals who are working smarter with AI. Learn how early adopters are using AI in their work and businesses.