Step 3.7 Flash
Neural Network
Step 3.7 Flash is a StepFun multimodal model for text, images, and video, available via API on BotHub.
Max answer length
(in tokens)
Context size
(in tokens)
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Step 3.7 Flash?
Step 3.7 Flash is a multimodal StepFun model based on the Mixture-of-Experts architecture: the language base has 196B parameters, but about 11B are activated per request, and the built-in visual encoder adds native understanding of images and video. Thanks to this scheme, the model remains fast while maintaining quality in parsing text and visual data. Via BotHub, it is accessible without a VPN or foreign card; payment is made with Russian cards in rubles based on actual usage, and unused tokens do not expire. With over 250 neural networks in one window, you can compare responses and switch models without rewriting the integration: a single OpenAI-compatible API works with one key, traffic is encrypted via AES-GCM, and companies have access to contracts, invoices, electronic document management, and an admin panel with limits. It is suitable for analyzing interface screenshots, diagrams, and blueprints, describing video content, building multimodal assistants via API, extracting data from scans and photos of documents, and creating scenarios where text and images are analyzed together.Frequently asked questions about Step 3.7 Flash
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
step-3.7-flash is a text model from StepFun with a 256,000 token context. It accepts text, documents, and images as input, reasons before answering, and calls functions. It is suitable for parsing large documents, coding and debugging, STEM tasks, and embedding into your own services via API.
In one response, the model outputs up to 230,400 tokens — enough for voluminous documentation, large code analysis, or an entire long report without manual stitching. And the 256,000 token context allows you to keep source materials and the result in one dialogue.
Yes, reasoning before answering is a stated capability of the model: before formulating a result, it breaks down the task step-by-step. This significantly helps in mathematics, logic, data analysis, and code debugging, where not only the result but the path to it is important.
Yes, the model supports function calling and structured JSON output, making it easy to connect to external tools, databases, and your own services. In BotHub, this works via a single OpenAI-compatible API — you can switch models without rewriting the integration.
Images and documents — yes, they can be sent along with text: analyze a screenshot, diagram, table, contract, or presentation. Audio and video are not marked in our data, so we will not promise such input — it is easier to check the required format directly in the interface.
We do not have data on prompt and response languages, so we will not make any claims. It is more reliable to test the model with your own text and compare it with others in the same window: in BotHub, it is available from Russia without a VPN or foreign card, with pay-as-you-go token billing.