selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

Alternatives to LFM3-Audio-2B

114 open-weight models that do the same job as LFM3-Audio-2B (2B omni), ranked by how close they are in task and size, then by popularity and how recent they are. Licence and VRAM are shown for each.

the original
LiquidAI logo

LFM3-Audio-2B

LiquidAI·Omni

≈2.2 GB

LFM3-Audio-2B is a compact, fully featured audio model that runs locally on laptops and edge devices. It takes text or audio as input and responds with text, speech, or both.

2B125K ctx lfm1.0
18.2KSept 2026ollama
--vram
any
--gpus
--quant
  1. ai-sage logo≈8.1 GB

    GigaChat Audio 10B is an audio-native LLM built on top of the GigaChat 3.1 Lightning text model. A Conformer speech encoder and a modality adapter feed audio embeddings directly into a Mixture-of-Experts decoder, so the model keeps the text quality of its base while adding speech understanding.

    10BA1.8B256K ctx MIT
    789Jul 2026
    • Same category
    • Larger (10B vs 2B)
    • Commercial use OK
    vs
  2. FunAudioLLM logo

    Fun-Audio-Chat-8B

    FunAudioLLM·Omni

    ≈7.2 GB

    Fun-Audio-Chat is a Large Audio Language Model built for natural, low-latency voice interactions. It introduces Dual-Resolution Speech Representations (an efficient 5Hz shared backbone + a 25Hz refined head) to cut compute while keeping high speech quality, and Core-Cocktail training to preserve…

    9.5B256K ctx Apache 2.0
    2.2KDec 2025
    • Same category
    • Larger (9.5B vs 2B)
    • Commercial use OK
    vs
  3. FlashLabs logo

    Chroma-4B

    FlashLabs·Omni

    ≈4.9 GB

    FlashLabs Chroma 1.0: A Real-Time End-to-End Spoken Dialogue Model with Personalized Voice Cloning

    5.9BApache 2.0
    46Nov 2025
    • Same category
    • Larger (5.9B vs 2B)
    • Commercial use OK
    vs
  4. XiaomiMiMo logo

    MiMo-Audio-7B-Instruct

    XiaomiMiMo·Omni

    ≈6.4 GB

    Existing audio language models typically rely on task-specific fine-tuning to accomplish particular audio tasks. In contrast, humans are able to generalize to new audio tasks with only a few examples or simple instructions.

    8B8K ctx MIT
    847Sept 2025
    • Same category
    • Larger (8B vs 2B)
    • Commercial use OK
    vs
  5. nvidia logo

    audio-flamingo-2

    nvidia·Omni

    ≈2.9 GB

    This repo contains the PyTorch implementation of Audio Flamingo 2: An Audio-Language Model with Long-Audio Understanding and Expert Reasoning Abilities. Audio Flamingo 2 achieves the state-of-the-art performance across over 20 benchmarks, with only a 3B parameter small language model.

    3BCustom licence
    0Feb 2025
    • Same category
    • Similar size (3B vs 2B)
    vs
  6. OpenMOSS-Team logo

    MOSS-Music-8B-Thinking

    OpenMOSS-Team·Omni

    ≈7.0 GB

    MOSS-Music is an open-source music understanding model from MOSI.AI, the OpenMOSS team, and Shanghai Innovation Institute.

    9.1B40K ctx Apache 2.0
    459Apr 2026gguf
    • Same category
    • Larger (9.1B vs 2B)
    • Commercial use OK
    vs
  7. XiaomiMiMo logo

    MiMo-Audio-7B-Base

    XiaomiMiMo·Omni

    ≈6.4 GB

    Existing audio language models typically rely on task-specific fine-tuning to accomplish particular audio tasks. In contrast, humans are able to generalize to new audio tasks with only a few examples or simple instructions.

    8B8K ctx MIT
    673Sept 2025
    • Same category
    • Larger (8B vs 2B)
    • Commercial use OK
    vs
  8. ByteDance-Seed logo

    Tar-7B

    ByteDance-Seed·Omni

    ≈6.1 GB

    Tar-7B is an open-weight omni model with 8.7B parameters from ByteDance-Seed, published on Hugging Face.

    8.7B32K ctx Apache 2.0
    1.7KJul 2025
    • Same category
    • Larger (8.7B vs 2B)
    • Commercial use OK
    vs
  9. naver-hyperclovax logo

    HyperCLOVAX-SEED-Omni-8B

    naver-hyperclovax·Omni

    ≈8.0 GB

    HyperCLOVA X SEED 8B Omni is a unified multimodal model that brings text, vision, and speech together, based on an auto-regressive Transformer architecture, enabling consistent multimodal understanding and generation.

    10.7B8K ctx hyperclovax
    2.6KDec 2025gguf
    • Same category
    • Larger (10.7B vs 2B)
    vs
  10. google logo

    gemma-3n-E4B

    google·Omni

    ≈5.7 GB

    Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. Gemma 3n models are designed for efficient execution on low-resource devices.

    7.8B32K ctx Gemma Terms of Use
    737Jun 2025gguf
    • Same category
    • Larger (7.8B vs 2B)
    vs
  11. OpenMOSS-Team logo

    MOVA-720p

    OpenMOSS-Team·Omni

    ? GB

    We introduce MOVA (MOSS Video and Audio), a foundation model designed to break the "silent era" of open-source video generation. Unlike cascaded pipelines that generate sound as an afterthought, MOVA synthesizes video and audio simultaneously for perfect alignment.

    ?BApache 2.0
    875Jan 2026
    • Same category
    • Commercial use OK
    vs
  12. bytedance-research logo

    Lance

    bytedance-research·Omni

    ? GB

    Lance is a 3B native unified multimodal model that supports image and video understanding, generation, and editing within a single framework.

    ?BApache 2.0
    381May 2026gguf
    • Same category
    • Commercial use OK
    vs
  13. kyutai logo≈6.5 GB

    A speech-to-speech model that reasons before it speaks. GLM-4-Voice-9B trained with reinforcement learning against a binary LLM judge on math word problems — no supervised finetuning stage, the reasoning behaviour is learned from the reward alone.

    9.5B8K ctx glm-4-voice-license
    107Aug 2026
    • Same category
    • Larger (9.5B vs 2B)
    vs
  14. Qwen logo

    Qwen2-Audio-7B

    Qwen·Omni

    ≈6.5 GB

    Qwen2-Audio is the new series of Qwen large audio-language models. Qwen2-Audio is capable of accepting various audio signal inputs and performing audio analysis or direct textual responses with regard to speech instructions. We introduce two distinct audio interaction modes:

    8.4BApache 2.0
    6.7KJul 2024gguf
    • Same category
    • Larger (8.4B vs 2B)
    • Commercial use OK
    vs
  15. NVIDIA Nemotron 3 Nano Omni is a multimodal large language model that unifies video, audio, image, and text understanding to support enterprise-grade Q&A, summarization, transcription, and document intelligence workflows.

    33BA3B256K ctx nvidia-open-model-agreement
    93.9KApr 2026gguf
    • Same category
    • Larger (33B vs 2B)
    • Commercial use OK
    vs
  16. Qwen logo≈22 GB

    Qwen3-Omni is the natively end-to-end multilingual omni-modal foundation models. It processes text, images, audio, and video, and delivers real-time streaming responses in both text and natural speech. We introduce several architectural upgrades to improve performance and efficiency. Key features:

    35.3BA3B64K ctx Apache 2.0
    340.5KSept 2025gguf
    • Same category
    • Larger (35.3B vs 2B)
    • Commercial use OK
    vs
  17. FreedomIntelligence logo

    Janus-4o-7B

    FreedomIntelligence·Omni

    ≈5.9 GB

    Janus-4o is a multimodal large language model (MLLM) capable of both text-to-image and text-and-image-to-image generation. It is fine-tuned from Janus-Pro-7B using the ShareGPT-4o-Image dataset to align Janus-Pro with GPT-4o image generation capabilities.

    7.4B16K ctx MIT
    123Jun 2025
    • Same category
    • Larger (7.4B vs 2B)
    • Commercial use OK
    vs
  18. sensenova logo

    SenseNova-U1.5-8B-MoT

    sensenova·Omni

    ≈12 GB

    SenseNova-U1.5-8B-MoT is our latest native unified multimodal checkpoint for more accurate, consistent, reliable, and aesthetically compelling visual creation.

    17.5B256K ctx Apache 2.0
    6.5KAug 2026gguf
    • Same category
    • Larger (17.5B vs 2B)
    • Commercial use OK
    vs
  19. Qwen logo≈20 GB

    Qwen3-Omni is the natively end-to-end multilingual omni-modal foundation models. It processes text, images, audio, and video, and delivers real-time streaming responses in both text and natural speech. We introduce several architectural upgrades to improve performance and efficiency. Key features:

    31.7BA3B64K ctx Apache 2.0
    150KSept 2025gguf
    • Same category
    • Larger (31.7B vs 2B)
    • Commercial use OK
    vs
  20. google logo≈0.6 GB

    Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants.

    78M128K ctx Apache 2.0
    30.3KApr 2026gguf
    • Same category
    • Smaller (78M vs 2B)
    • Commercial use OK
    vs
  21. nvidia logo

    audio-flamingo-3

    nvidia·Omni

    ? GB

    Audio Flamingo 3: Advancing Audio Intelligence with Fully Open Large Audio-Language Models

    ?BCustom licence
    708Jul 2025
    • Same category
    vs
  22. google logo≈0.6 GB

    Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants.

    79M128K ctx Apache 2.0
    23.9KApr 2026gguf
    • Same category
    • Smaller (79M vs 2B)
    • Commercial use OK
    vs
  23. inclusionAI logo

    Ming-flash-omni-2.0

    inclusionAI·Omni

    ≈63 GB

    The newly released Ming-flash-omni 2.0 leverages the Ling-2.0 architecture—a Mixture-of-Experts (MoE) framework comprising 100B total and 6B active parameters.

    104BA7.5B32K ctx MIT
    31.1KFeb 2026
    • Same category
    • Larger (104B vs 2B)
    • Commercial use OK
    vs
  24. BAAI logo

    Emu3-Gen

    BAAI·Omni

    ≈6.5 GB

    We introduce Emu3, a new suite of state-of-the-art multimodal models trained solely with next-token prediction ! By tokenizing images, text, and videos into a discrete space, we train a single transformer from scratch on a mixture of multimodal sequences.

    8.5B9K ctx Apache 2.0
    755Sept 2024
    • Same category
    • Larger (8.5B vs 2B)
    • Commercial use OK
    vs