Skip to main content

StepFun Provider Guide

StepFun is a Tier-2 catalog provider: its integration is one JSON file (src/lib/providers/catalog/stepfun.json) rather than hand-written code.

Verification status: this entry is docs-verified only, not yet live-verified. StepFun's GET https://api.stepfun.ai/v1/models needs a key (it answers HTTP 401 without one), so the five model ids come from the vendor's public pricing page and the roster has not been checked. No account was created and no API key was used to build this entry. evidence.liveMatrix is null until someone runs the live capability matrix with a real key (see Verification status below).

Key Facts​

  • Provider id: stepfun
  • Protocol: OpenAI-compatible (/chat/completions) — the vendor's migration page says "Stepfun models are compatible with the OpenAI API spec."
  • Base URL: https://api.stepfun.ai/v1
  • Step Plan: the vendor's Step Plan overview lists a separate OpenAI SDK base URL, https://api.stepfun.ai/step_plan/v1; this entry's default is the standard endpoint above
  • Default model: step-5-preview
  • Models in catalog: 5 (ids taken from the vendor's pricing page; roster not checked)
  • Streaming: supported — the Chat Completions page documents stream and chat.completion.chunk events ending in data: [DONE]
  • Tool calling: model-dependent — the Tool Call page lists step-5-preview, step-3.7-flash, step-3.5-flash and step-3.5-flash-2603 among the models that support Toolcall requests
  • Tools while streaming: supported — the streaming chunk format on the Chat Completions page includes a tool_calls field in delta
  • Structured output: supported — the Chat Completions page documents response_format { "type": "json_object" }, and the JSON Mode page says the json_schema approach applies to step-5-preview, step-3.7-flash, step-3.5-flash and step-3.5-flash-2603
  • Structured output + tools together: not declared (false) — no combined probe was possible without credentials
  • Embeddings: not declared
  • Thinking: declared (true) — the Chat Completions page documents a reasoning_effort request parameter and a reasoning response field
  • Billing: recorded as no-free-tier because the catalog schema has no unknown value; that is not a vendor statement. The vendor's own billing wording is quoted under Billing below
  • Key format: none declared

Quick Start​

1. Get an API key​

  1. Visit: https://platform.stepfun.ai/interface-key and sign in - the Quickstart (https://platform.stepfun.ai/docs/en/quickstart/overview.md) says "Sign in to the StepFun Open Platform and create and copy an API key on the Interface Keys page."
  2. Billing as the vendor states it: the pricing overview (https://platform.stepfun.ai/docs/en/guides/pricing/intro.md) says fees are deducted "drawing from free credit first, then paid balance", and the Error Codes page (https://platform.stepfun.ai/docs/en/api-reference/error-codes.md) lists HTTP 402 as "Insufficient balance" with the resolution "Add funds to your account."
  3. Set STEPFUN_API_KEY in your .env file

2. Configure​

export STEPFUN_API_KEY=your-api-key
export STEPFUN_MODEL=step-5-preview   # optional — overrides the default model
export STEPFUN_BASE_URL=https://api.stepfun.ai/v1   # optional — proxy or self-hosted gateway

3. Use it​

import { NeuroLink } from "@juspay/neurolink";

const neurolink = new NeuroLink();

const result = await neurolink.generate({
  input: { text: "Explain context windows in one paragraph." },
  provider: "stepfun",
  model: "step-5-preview",
});

console.log(result.content);
# CLI
npx @juspay/neurolink generate "Hello" --provider stepfun

Per-request credentials are passed the same way as for the other catalog providers:

await neurolink.generate({
  input: { text: "Hello" },
  provider: "stepfun",
  credentials: { stepfun: { apiKey: process.env.STEPFUN_API_KEY } },
});

Billing​

The pricing overview (pricing/intro) says fees are deducted "drawing from free credit first, then paid balance". The Error Codes page lists HTTP 402 as "Insufficient balance" with the resolution "Add funds to your account." The pricing details page says pay-as-you-go calls to the standard Open Platform API "are rate limited by tier according to the cumulative top-up amount on your account", and for individual accounts states "Individual accounts must complete identity verification first." Its individual-account tiers:

User tierCumulative top-up amountConcurrencyRPMTPM
V0Under $155100500,000
V1$15 to $69204002,000,000
V2$70 to $299306003,000,000
V3$300 to $1,499408004,000,000
V4$1,500 and above1302,60013,000,000

The entry records no-free-tier because the catalog schema requires one of free-tier, free-with-card or no-free-tier; the vendor pages quoted above are the billing statement.

Models​

ModelContextMax outputVision$/M in · out (cache hit)Notes
step-5-preview ⭐1M64kyes$1.00 / $2.70 ($0.05)NeuroLink default. "Step 5 Preview is StepFun's flagship model for agentic work." The Chat Completions page states "The examples below use step-5-preview by default."
step-3.7-flash256K—yes$0.20 / $1.15 ($0.04)Fallback. "StepFun's flagship multimodal reasoning model."
step-3.5-flash256K—no$0.10 / $0.30 ($0.02)Fallback. "StepFun's flagship language reasoning model"; the Quickstart table lists Input capabilities "Text only".
step-3.5-flash-2603256K—no$0.10 / $0.30 ($0.02)Fallback. "Agent-optimized version: tuned from step-3.5-flash for high-frequency agent scenarios."
step-1o-turbo-vision32K—yes$0.36 / $1.15 ($0.07)Fallback. Listed under "Pricing for Vision Models"; the welcome-page card says "Designed for social content creation, AI assistants, and image and video understanding."

Context, max output and prices are the vendor's figures from these pages (retrieved 2026-09-29):

  • step-5-preview: the model page gives Context window "1M tokens", Input types "Text, images, and video" and Maximum output "64k tokens".
  • step-3.7-flash: the model page states a context length of 256K tokens; the Quickstart table lists Input capabilities "Text, images, and video".
  • step-3.5-flash: the model page states a context length of 256K tokens.
  • step-3.5-flash-2603: the same model page describes it under Model variants; the welcome page card titled "Step 3.5 Flash 2603" shows the metric "Max context" (value 256K).
  • step-1o-turbo-vision: the welcome-page card titled "Step-1o Turbo Vision" shows the metric "Max context" (value 32K); the Model Retirement and Migration page lists it in the "Recommended replacement" column for step-1-8k, step-1-32k, step-1v-8k, step-1v-32k, step-2-mini, step-1o-vision-32k and step-2-16k.
  • Prices for the five models: the pricing details page (per 1M tokens, input cache miss / input cache hit / output).

The catalog stores the labels as decimal numbers (1M as 1,000,000, 256K as 256,000, 32K as 32,000, 64k as 64,000). models.defaultContextWindow (32,000) and models.defaultMaxOutputTokens (16,384) are conservative placeholders for model ids outside the catalog; the vendor does not publish a general default.

Vision: vision: true for step-5-preview and step-3.7-flash follows the Input capabilities "Text, images, and video" in the Quickstart table, and for step-1o-turbo-vision follows its "Pricing for Vision Models" listing and card text. step-3.5-flash and step-3.5-flash-2603 are vision: false; the welcome-page card for step-3.5-flash-2603 carries the category "Reasoning / text".

Status: step-5-preview is marked preview because the vendor names it "Step 5 Preview".

Fallback order when the default is unavailable: step-3.7-flash → step-3.5-flash → step-3.5-flash-2603 → step-1o-turbo-vision. models.fallbackModelName is set to step-3.7-flash, which the Quickstart table lists with the same Input capabilities ("Text, images, and video") as the default. models.visionModel is step-5-preview.

Verification status​

Tier-2 onboarding requires evidence before a provider is accepted, and pnpm run verify:provider-onboarding gates it in CI. This is what the catalog currently records for StepFun — docs-verified only, not live-verified:

ProbeResult
Rosterunauthenticated GET /v1/models answers HTTP 401 without a key (2026-09-29), so the roster has not been checked; model ids come from the vendor's pricing page (retrieved 2026-09-29)
Billingvendor wording quoted under Billing: "drawing from free credit first, then paid balance"; HTTP 402 "Insufficient balance"; tiered rate limits by cumulative top-up amount, 2026-09-29
Auth-failure shapeunauthenticated GET /v1/models and GET /v1/chat/completions each answered HTTP 401 with the body {"error":{"message":"Incorrect API key provided","type":"invalid_api_key"}} on 2026-09-29 (the body has no error.code field; authProbe.code records error.type). errorRules match 401 plus 402, 429, 451 and 503 from the Error Codes page
Tools / structured outputdocumented on the Tool Call and JSON Mode pages; neither was exercised live, and no combined tools+schema request was sent — structuredOutputWithTools stays false. No POST request was sent, so the request fields NeuroLink's client can add (stream_options, tool_choice, a json_schema response_format) have not been exercised against StepFun
Live capability sweepnot run. evidence.liveMatrix is null. Before treating this provider as production-ready, run npx tsx test/continuous-test-suite-provider-matrix.ts --provider=stepfun with a real key and record the result.

Do not treat this entry as equivalent to a live-verified Tier-2 provider (e.g. FriendliAI, Novita AI) until that live matrix has been run and evidence.liveMatrix is filled in.

Troubleshooting​

These rows restate the vendor's Error Codes page (retrieved 2026-09-29).

SymptomCause (vendor wording)Fix (vendor wording)
HTTP 400"Request parameters are invalid.""Check the docs and pass the parameters correctly."
HTTP 401"Invalid authentication""Make sure you are using the correct API key."
HTTP 402"Insufficient balance""Add funds to your account."
HTTP 404"Incorrect request path""Update the request path per the docs."
HTTP 429"Resource or rate limit exceeded, usually because requests are too frequent""Retry after a short delay."
HTTP 451"The request or response content failed review""Modify the request and try again."
HTTP 500"Server-side issue on our end""Retry shortly; contact us if it persists."
HTTP 503"Server is currently overloaded""Retry after a short delay."

See also​