Contents Menu Expand Light mode Dark mode Auto light/dark, in light mode Auto light/dark, in dark mode Skip to content
mobius
mobius

User Guide

  • Getting Started
  • CLI Reference
  • Module Architecture Guide
  • Model Catalog
  • Supported Models
    • DFlashDraftModel
    • Eagle3DraftModel
    • Eagle3LlamaForCausalLM
    • Eagle3Speculator
    • Gemma4AssistantForCausalLM
    • Gemma4UnifiedAssistantForCausalLM
    • LlamaForCausalLMEagle3
    • Qwen35MtpModel
    • albert
    • apertus
    • arcee
    • arctic
    • aya_vision
    • baichuan
    • bamba
    • bart
    • beit
    • bert
    • bigbird_pegasus
    • biogpt
    • blenderbot
    • blenderbot-small
    • blip
    • blip-2
    • bloom
    • bros
    • camembert
    • chameleon
    • chatglm
    • clip_text_model
    • clip_vision_model
    • code_llama
    • codegen
    • codegen2
    • cohere
    • cohere2
    • cohere2_vision
    • command_r
    • csm
    • ctrl
    • cvt
    • data2vec-audio
    • data2vec-text
    • data2vec-vision
    • dbrx
    • deberta
    • deberta-v2
    • deepseek_v2
    • deepseek_v2_moe
    • deepseek_v3
    • deepseek_v4
    • deepseek_vl
    • deepseek_vl_hybrid
    • deepseek_vl_v2
    • deit
    • depth_anything
    • diffllama
    • dinov2
    • dinov2_with_registers
    • dinov3_vit
    • distilbert
    • doge
    • dots1
    • electra
    • ernie
    • ernie4_5
    • ernie4_5_moe
    • ernie_m
    • esm
    • evolla
    • exaone
    • exaone4
    • falcon
    • falcon_h1
    • falcon_mamba
    • fastconformer_rnnt
    • flaubert
    • flex_olmo
    • florence2
    • fsmt
    • fun_asr
    • fuyu
    • gemma
    • gemma2
    • gemma3
    • gemma3_text
    • gemma3n
    • gemma3n_text
    • gemma4
    • gemma4_assistant
    • gemma4_text
    • gemma4_unified
    • gemma4_unified_assistant
    • gemma4_unified_text
    • glm
    • glm4
    • glm4_moe
    • glm4v
    • glm4v_moe
    • glm4v_moe_text
    • glm4v_text
    • got_ocr2
    • gpt-sw3
    • gpt2
    • gpt_bigcode
    • gpt_neo
    • gpt_neox
    • gpt_neox_japanese
    • gpt_oss
    • gptj
    • granite
    • granitemoe
    • granitemoehybrid
    • granitemoeshared
    • helium
    • hiera
    • hubert
    • hunyuan_v1_dense
    • hunyuan_v1_moe
    • hunyuan_vl_mot
    • ibert
    • idefics2
    • idefics3
    • ijepa
    • imagegpt
    • instructblip
    • instructblipvideo
    • internlm2
    • internvl
    • internvl2
    • internvl_chat
    • jamba
    • janus
    • jetmoe
    • layoutlm
    • layoutlmv2
    • layoutlmv3
    • led
    • lilt
    • llada
    • llama
    • llama4_text
    • llava
    • llava_next
    • llava_next_video
    • llava_onevision
    • longcat_flash
    • longt5
    • m2m_100
    • mamba
    • mamba2
    • marian
    • markuplm
    • mbart
    • mctct
    • mega
    • megatron-bert
    • minicpm
    • minicpm3
    • minimax
    • ministral
    • ministral3
    • mistral
    • mistral3
    • mixtral
    • mllama
    • mms
    • mobilebert
    • mobilevit
    • mobilevitv2
    • modernbert
    • modernbert-decoder
    • molmo
    • mpnet
    • mpt
    • mra
    • mt5
    • musicgen
    • mvp
    • nanochat
    • nemotron
    • nemotron_h
    • nezha
    • nllb-moe
    • nllb_moe
    • nystromformer
    • olmo
    • olmo2
    • olmo3
    • olmoe
    • open-llama
    • openai-gpt
    • openelm
    • opt
    • ovis2
    • paligemma
    • pegasus
    • pegasus_x
    • persimmon
    • phi
    • phi3
    • phi3_v
    • phi3small
    • phi4-siglip
    • phi4_multimodal
    • phi4mm
    • phimoe
    • pixtral
    • plbart
    • prophetnet
    • pvt
    • pvt_v2
    • qdqbert
    • qwen
    • qwen2
    • qwen2_5_vl
    • qwen2_5_vl_text
    • qwen2_moe
    • qwen2_vl
    • qwen2_vl_text
    • qwen3
    • qwen3_5
    • qwen3_5_moe
    • qwen3_5_moe_vl
    • qwen3_5_text
    • qwen3_5_vl
    • qwen3_5_vl_text
    • qwen3_asr
    • qwen3_forced_aligner
    • qwen3_moe
    • qwen3_next
    • qwen3_omni_moe
    • qwen3_tts
    • qwen3_tts_tokenizer_12hz
    • qwen3_vl
    • qwen3_vl_moe
    • qwen3_vl_single
    • qwen3_vl_text
    • rembert
    • roberta
    • roberta-prelayernorm
    • roc_bert
    • roformer
    • sam2
    • seamless_m4t
    • seamless_m4t_v2
    • seed_oss
    • segformer
    • sensevoice_small
    • sew
    • sew-d
    • shieldgemma2
    • siglip
    • siglip2
    • siglip2_vision_model
    • siglip_vision_model
    • smollm3
    • smolvlm
    • solar_open
    • speecht5
    • splinter
    • squeezebert
    • stablelm
    • starcoder2
    • swin
    • swin2sr
    • swinv2
    • switch_transformers
    • t5
    • trocr
    • umt5
    • unispeech
    • unispeech-sat
    • video_llava
    • vipllava
    • vit
    • vit_hybrid
    • vit_mae
    • vit_msn
    • voxtral_encoder
    • wav2vec2
    • wav2vec2-bert
    • wav2vec2-conformer
    • wavlm
    • whisper
    • xglm
    • xlm
    • xlm-prophetnet
    • xlm-roberta
    • xlm-roberta-xl
    • xlnet
    • xmod
    • yi
    • yolos
    • yoso
    • youtu
    • zamba
    • zamba2

Execution Providers

  • EP-Aware Building: Quick Start
  • Execution Provider (EP) Aware Building

Reference

  • API Reference
    • build()
    • build_from_module()
    • build_from_gguf()
    • apply_weights()
    • ModelPackage
    • BaseModelConfig
    • ModelRegistry
  • Feature Flags

Design

  • Design Documents
    • System Architecture
    • Config System Redesign: mobius
    • DeepSeek-V4-Flash export design
    • GGUF Support Proposal for mobius
    • Golden File Infrastructure Design — L4 & L5 Testing
    • Multi-Tier Testing Strategy
    • Performance Evaluation & Regression Testing Infrastructure
    • Phi-4-Multimodal-Instruct Architecture Breakdown
    • Phi4MM Four-Model Split Design
    • Phi4MM ORT-GenAI Integration Spec
    • ONNX GenAI Models - Test Infrastructure Comprehensive Map

Contributing

  • Strategy: AI-Assisted Model Support

Research

  • Testing Strategy Analysis
Back to top
View this page

API Reference¶

Created: 2026-03-18 · Last updated: 2026-04-08

Public API documentation for mobius.

Core Functions¶

  • build() — Build from a HuggingFace model ID

  • build_from_module() — Build from a module instance

  • build_from_gguf() — Build from a GGUF file

  • apply_weights() — Apply weights to a built model

Core Classes¶

  • ModelPackage — Collection of named ONNX models

  • BaseModelConfig — Base configuration class

  • ModelRegistry — Architecture registry

Next
build()
Previous
Execution Provider (EP) Aware Building
Copyright © 2026, Microsoft Corporation
Made with Sphinx and @pradyunsg's Furo
On this page
  • API Reference
    • Core Functions
    • Core Classes