StepFun

Step 5 Preview

Generally availableOpen weights1M context

Summary

Step 5 Preview is an open-weight language model from StepFun, released on 16 Sept 2026. It accepts text and images and generates text, with a 1M-token context window and up to 1M output tokens per response. As of 18 Nov 2026, the lowest listed API price is $1 per million input tokens and $2.70 per million output tokens (Novita AI), across 4 providers we track.

StepFun flash model for efficient multimodal reasoning, coding, and tool use

Last verified 18 Nov 2026. Every figure on this page links to its source.

Specifications

DeveloperStepFun
Release date16 Sept 2026 [models.dev]
StatusGenerally available [models.dev]
WeightsOpen weights [Hugging Face]
Parameters604B [Hugging Face]
Context window1M (1,000,000 tokens) [models.dev]
Max output1M tokens [models.dev]
InputText and images [models.dev]
OutputText [models.dev]
Reasoning modeNo [models.dev]
Tool callingNo [models.dev]
Structured outputYes [models.dev]
Hugging Facestepfun-ai/Step-5-Preview [Hugging Face]

Step 5 Preview API pricing by provider

USD per million tokens. Sorted by input price.

ProviderInputOutputCached inputContextSource
NVIDIA NIM
stepfun-ai/step-5-preview
FreeFree–1Mmodels.dev, 18 Nov 2026
Novita AI
stepfun/step-5-preview
$1$2.70$0.101Mmodels.dev, 18 Nov 2026
OpenRouter
stepfun/step-5-preview
$1$2.70$0.101MOpenRouter, 18 Nov 2026
Vercel AI Gateway
stepfun/step-5-preview
$1$2.70$0.101Mmodels.dev, 18 Nov 2026

Running Step 5 Preview locally

Estimated memory, from the parameter count and the architecture in config.json on Hugging Face. Includes about 0.8 GB of runtime overhead.

QuantizationWeightsTotal, 8K contextTotal, 32K context
Q4_K_M (4-bit)366 GB378 GB406 GB
Q8_0 (8-bit)642 GB654 GB682 GB
FP16 / BF16 (16-bit)1209 GB1221 GB1248 GB

At 4-bit with an 8K conversation, Step 5 Preview needs about 378 GB, more than any single consumer graphics card; it needs a unified-memory machine, a multi-GPU setup or a split between graphics card and system RAM. Check your own hardware, or see Best computer for local AI on our sister site AIPCs.

Frequently asked questions

What is the context window of Step 5 Preview?

Step 5 Preview has a context window of 1,000,000 tokens (1M) and can generate up to 1,000,000 tokens in a single response, according to models.dev.

How much does the Step 5 Preview API cost?

Step 5 Preview is listed from $1 per million input tokens and $2.70 per million output tokens (Novita AI). Prices last verified 18 Nov 2026.

Is Step 5 Preview open source?

Step 5 Preview is an open-weight model: its weights are published on Hugging Face as stepfun-ai/Step-5-Preview. Check the license terms before commercial use.

When was Step 5 Preview released?

StepFun released Step 5 Preview on 16 Sept 2026, according to models.dev.

Which providers offer Step 5 Preview?

We track Step 5 Preview on 4 providers: NVIDIA NIM, Novita AI, OpenRouter, Vercel AI Gateway.

How much VRAM does Step 5 Preview need?

About 378 GB at 4-bit (Q4_K_M) with an 8K-token conversation, and about 654 GB at 8-bit. That is more than any single consumer graphics card holds. These are estimates from the model's published parameter count and architecture.

Similar models from other labs

Comparable on output type, weights, release date, price and context window. Prices are the lowest listed, USD per million tokens.

ModelLabReleasedContextInput $/1MOutput $/1MWeights
Pareto 26.10 PreviewUnbiased1 Oct 20261M$0.80$3.20–
Tencent Hy4 PreviewTencent (Hunyuan)28 Aug 20261M$0.83$2.50Open
Macaron V1 VentiMindlab21 Jul 20261M$0.82$2.48Open
Muse Spark OpenMeta15 Oct 20261M$1.25$4.25Open
GLM 5.3 PrimeZhipu AI (Z.ai)23 Sept 20261M$1.68$5.28–
ERNIE-Light-DT-1.0Baidu10 Nov 20261M$1.13$3.94–

Change history

  • Context window changed from 1M to 1M
  • Tool calling changed from yes to no
  • Reasoning mode changed from yes to no
  • Structured output changed from no to yes
  • Max output changed from 1M to 1M
  • Max output changed from 64K to 1M
  • Input modalities changed from text, image, video to text, image
  • Context window changed from 1M to 1M
  • Structured output changed from yes to no
  • Input modalities changed from text, image to text, image, video
  • Structured output changed from no to yes
  • Max output changed from 1M to 64K
  • Context window changed from 1M to 1M
  • Tool calling changed from no to yes
  • Structured output changed from yes to no
  • Reasoning mode changed from no to yes
  • Structured output changed from no to yes
  • Tool calling changed from yes to no
  • Reasoning mode changed from yes to no
  • Structured output changed from yes to no
  • Reasoning mode changed from no to yes
  • Tool calling changed from no to yes
  • Structured output changed from no to yes
  • Tool calling changed from yes to no
  • Reasoning mode changed from yes to no

Sources