Meet the team that turns your ideas into work.

Discover EightBots
EIGHTBOTS
Contact salesSign up
AI provider

StepFun

StepFun develops the Step model families for language, visual tasks and speech. Its catalog includes hosted APIs and published research weights.

Explore models

Models & tools

  • A multimodal mixture-of-experts model with 600B total and 27B active parameters for text, image and video tasks.

    Launch date not confirmedPreviewAPI · Downloadable · Provider product

  • A visual reasoning model with a 196B language backbone and 1.8B vision encoder for images, video, code and tool calls.

    Launch date not confirmedAPI · Downloadable

  • Step 3.5 Flash is a text reasoning model for code and tool use.

    Launch date not confirmedAPI · Downloadable

  • An image-editing model that adds reasoning and reflection to the Step1X-Edit pipeline.

    Launch date Downloadable

  • An instruction-based image-editing model that also supports text-to-image generation.

    Launch date Downloadable

  • A two-stage 3D generation model with separate geometry and texture models.

    Launch date Downloadable

  • Step1X-Edit combines a multimodal language model that interprets edits with a diffusion transformer that generates the image.

    Launch date Downloadable

  • A video-generation model that animates a starting image from a text instruction.

    Launch date Downloadable
  • Step-Video T2V

    StepFun · Step-Video

    Step-Video T2V is a 30B text-to-video model that can generate up to 204 frames.

    Launch date Downloadable

  • Step-Video T2V Turbo is the inference-step-distilled version of Step-Video T2V.

    Launch date Downloadable

  • GELab-Zero 4B Preview is StepFun's GUI agent release based on Qwen3-VL-4B-Instruct.

    Launch date not confirmedPreviewDownloadable

  • GOT-OCR 2.0 is a 580M-parameter OCR model from joint research involving StepFun, UCAS, Tsinghua University and Megvii.

    Launch date not confirmedDownloadable

  • An image-generation model combining a 14B autoregressive backbone with a 157M flow-matching head.

    Launch date not confirmedDownloadable

  • NextStep 1 Large Edit adapts the NextStep image pipeline to an input image and an editing instruction.

    Launch date not confirmedDownloadable

  • An image-generation pretraining checkpoint combining a 14B autoregressive backbone with a 157M flow-matching head.

    Launch date not confirmedResearch releaseDownloadable

  • An autoregressive image-generation model trained with flow-based reinforcement learning.

    Launch date not confirmedDownloadable

  • A pretraining-stage checkpoint of the NextStep 1.1 autoregressive image-generation model.

    Launch date not confirmedResearch releaseDownloadable

  • PaCoRe 8B is a reasoning model trained to combine parallel solution paths.

    Launch date not confirmedResearch releaseDownloadable

  • A Qwen2.5-32B derivative trained for multi-turn reasoning between simulated experts.

    Launch date not confirmedResearch releaseDownloadable

  • A Qwen3-8B-Base derivative with reasoning-oriented post-training, used as the initial PaCoRe checkpoint.

    Launch date not confirmedResearch releaseDownloadable
  • Step 3

    StepFun · Step 3

    Step 3 is a 321B-parameter visual reasoning model with 38B active parameters.

    Launch date not confirmedDownloadable

  • A pretrained sparse language-model checkpoint from the Step 3.5 Flash family.

    Launch date not confirmedResearch releaseDownloadable

  • A midtraining-stage checkpoint of the Step 3.5 Flash sparse language model.

    Launch date not confirmedResearch releaseDownloadable
  • Step ASR

    StepFun · Step ASR

    Step ASR supports Chinese and English speech recognition.

    Launch date not confirmedAPI

  • A speech-recognition model with separate file-transcription and live-streaming variants.

    Launch date not confirmedDeprecated API; check access detailsAPI
  • Step TTS 2

    StepFun · Step TTS

    A speech-synthesis model that uses next-token prediction without separate speaker or emotion embeddings.

    Launch date not confirmedAPI

  • Step TTS Mini is an earlier hosted speech synthesis model with voice cloning and emotion or style controls.

    Launch date not confirmedAPI

  • Step-1o Audio is StepFun's first-generation end-to-end voice model.

    Launch date not confirmedAPI

  • A visual-understanding model that accepts text, images and video and returns text.

    Launch date not confirmedAPI
  • Step-Audio 2

    StepFun · StepAudio 2

    Step-Audio 2 understands speech, music, environmental sound and emotion, with tools and search.

    Launch date not confirmedAPI · Provider product

  • A speech-dialogue model supporting sound understanding, tool calls and retrieval from text and acoustic material.

    Launch date not confirmedAPI · Downloadable

  • A base audio-language checkpoint in the Step-Audio 2 framework before dialogue tuning.

    Launch date not confirmedResearch releaseDownloadable

  • A Step-Audio 2 Mini checkpoint for audio understanding and spoken dialogue.

    Launch date not confirmedResearch releaseDownloadable

  • A 130B audio-language model that generates spoken answers directly from audio input.

    Launch date not confirmedResearch releaseDownloadable

  • Step-Audio Chat is the 130B multimodal language component of Step-Audio.

    Launch date not confirmedDownloadable
  • Step-Audio EditX

    StepFun · Step-Audio EditX

    Step-Audio EditX is a 3B audio model trained with reinforcement learning for iterative changes to emotion, speaking style and vocal effects.

    Launch date not confirmedDownloadable
  • Step-Audio R1

    StepFun · Step-Audio R

    An audio-reasoning model trained with Modality-Grounded Reasoning Distillation to retain acoustic evidence.

    Launch date not confirmedRestricted accessDownloadable
  • Step-Audio R1.1

    StepFun · Step-Audio R

    An audio model that separates higher-level reasoning from speech generation through a dual-brain design.

    Launch date not confirmedDownloadable
  • Step-Audio R1.5

    StepFun · Step-Audio R

    Step-Audio R1.5 adds human-feedback training for natural dialogue while reasoning and speech proceed together.

    Launch date not confirmedAPI · Provider product

  • Step-Audio TTS 3B uses a dual-codebook language-model design for speech synthesis.

    Launch date not confirmedDownloadable

  • Step3-VL-10B combines a 1.8B PE-lang visual encoder with a Qwen3-8B decoder.

    Launch date not confirmedDownloadable

  • A visual-language base model combining a Perception Encoder with Qwen3-8B before chat post-training.

    Launch date not confirmedResearch releaseDownloadable

  • StepAudio 2 ASR Pro is the 32B recognition model in StepFun's audio catalog.

    Launch date not confirmedAPI
  • StepAudio 2.5 ASR

    StepFun · StepAudio 2.5

    StepAudio 2.5 ASR is a 4B speech recognition model that uses multi-token prediction.

    Launch date not confirmedAPI

  • StepAudio 2.5 Realtime accepts and returns speech over WebSocket.

    Launch date not confirmedAPI · Provider product
  • StepAudio 2.5 TTS

    StepFun · StepAudio 2.5

    StepAudio 2.5 TTS uses overall context and inline directions to control a voice performance.

    Launch date not confirmedAPI · Provider product

  • A speech-recognition model combining acoustic modeling with language-model context for Chinese, English and mixed-language recordings.

    Launch date not confirmedAPI · Provider product
  • StepAudio 3 Gen

    StepFun · StepAudio 3

    StepAudio 3 Gen creates audio from role descriptions, scripts and natural-language instructions.

    Launch date not confirmedPreviewAPI · Provider product

  • StepAudio 3 Music supports full songs, instrumental music and accompaniment for an isolated vocal.

    Launch date not confirmedPreviewAPI · Provider product

  • StepAudio 3 Realtime combines audio understanding, turn control, reasoning and tool use.

    Launch date not confirmedPreviewAPI · Provider product
  • StepAudio 3 TTS

    StepFun · StepAudio 3

    StepAudio 3 TTS produces speech from text and reference audio.

    Launch date not confirmedAPI · Provider product
  • StepFun Formalizer 32B

    StepFun · StepFun Formalizer

    A DeepSeek-R1-Distill-Qwen-32B-based model that converts mathematical text into Lean 4 statements.

    Launch date not confirmedResearch releaseDownloadable
  • StepFun Formalizer 7B

    StepFun · StepFun Formalizer

    StepFun Formalizer 7B converts natural-language mathematical problems into formal Lean 4 statements.

    Launch date not confirmedResearch releaseDownloadable

  • A DeepSeek-R1-Distill-Qwen-32B-based model for iterative Lean 4 proof search.

    Launch date not confirmedResearch releaseDownloadable

  • StepFun Prover Preview 7B is a theorem-proving model derived from DeepSeek-R1-Distill-Qwen-7B.

    Launch date not confirmedResearch releaseDownloadable