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. inclusionAI logo

    LLaDA2.0-Uni

    inclusionAI·Omni

    ≈10 GB

    LLaDA2.0-Uni: Unifying Multimodal Understanding and Generation with Diffusion Large Language Model

    16.3BA1.5B8K ctx Apache 2.0
    2.2KApr 2026
    • Same category
    • Larger (16.3B vs 2B)
    • Commercial use OK
    vs
  2. ByteDance-Seed logo

    BAGEL-7B-MoT

    ByteDance-Seed·Omni

    ≈9.7 GB

    We present BAGEL, an open‑source multimodal foundation model with 7B active parameters (14B total) trained on large‑scale interleaved multimodal data.

    14.7B32K ctx Apache 2.0
    9.1KMay 2025gguf
    • Same category
    • Larger (14.7B vs 2B)
    • Commercial use OK
    vs
  3. microsoft logo

    OmniParser

    microsoft·Omni

    ? GB

    OmniParser is a general screen parsing tool, which interprets/converts UI screenshot to structured format, to improve existing LLM based UI agent.

    ?BMIT
    1.6KOct 2024
    • Same category
    • Commercial use OK
    vs
  4. nvidia logo

    NLD-Uni

    nvidia·Omni

    ≈12 GB

    NLD-Uni is a unified multimodal model for image understanding and image generation. It combines a diffusion language model with specialized visual components and a mixture-of-transformers design.

    18.3B128K ctx Unknown licence
    1.3KOct 2026
    • Same category
    • Larger (18.3B vs 2B)
    • Newer release
    vs
  5. NVIDIA-NemotronLabs-AI-for-Media-Sports-Tennis is a fine-tuned multimodal large language model derived from Nemotron 3 Nano Omni. It was further trained on an in-house NVIDIA tennis dataset of manually labeled point-level video clips to specialize in tennis match understanding.

    33B256K ctx openmdw-1.1
    2.4KSept 2026gguf
    • Same category
    • Larger (33B vs 2B)
    • Commercial use OK
    vs
  6. BAAI logo

    Emu3-Stage1

    BAAI·Omni

    ≈6.2 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.

    8.5B5K ctx Apache 2.0
    199Oct 2024
    • Same category
    • Larger (8.5B vs 2B)
    • Commercial use OK
    vs
  7. Qwen logo

    Qwen-Audio

    Qwen·Omni

    ≈6.5 GB

    Qwen-Audio (Qwen Large Audio Language Model) is the multimodal version of the large model series, Qwen (abbr. Tongyi Qianwen), proposed by Alibaba Cloud. Qwen-Audio accepts diverse audio (human speech, natural sound, music and song) and text as inputs, outputs text.

    8.4B2K ctx Unknown licence
    1.6KNov 2023
    • Same category
    • Larger (8.4B vs 2B)
    vs
  8. ServiceNow-AI logo

    Apriel-Voice-27B-preview

    ServiceNow-AI·Omni

    ≈19 GB

    An open model for speech, reasoning, and tool use from ServiceNow

    27.2B256K ctx MIT
    1.5KOct 2026
    • Same category
    • Larger (27.2B vs 2B)
    • Commercial use OK
    • Newer release
    vs
  9. NVIDIA-NemotronLabs-AI-for-Media-Sports-Basketball is a fine-tuned multimodal large language model derived from Nemotron 3 Nano Omni.

    33B256K ctx openmdw-1.1
    1.2KSept 2026
    • Same category
    • Larger (33B vs 2B)
    • Commercial use OK
    • Newer release
    vs
  10. sarvamai logo

    shuka-1

    sarvamai·Omni

    ≈6.7 GB

    Shuka v1 is a language model which natively understands audio in Indic languages. It is an encoder-decoder model built by combining two models: - Our state-of-the-art, in-house, audio encoder: Saaras v1 - Meta’s Llama3-8B-Instruct as the decoder

    8.7BLlama 3 Community License
    238Aug 2024
    • Same category
    • Larger (8.7B vs 2B)
    vs
  11. KRAFTON logo≈22 GB

    A.X K2 Raon-Speech is a bilingual English/Korean speech language model with approximately 21.2B total parameters and 3.5B active parameters.

    21.2BA3B128K ctx CC BY-NC 4.0
    1.4KJul 2026
    • Same category
    • Larger (21.2B vs 2B)
    vs
  12. sensenova logo

    SenseNova-U1-8B-MoT

    sensenova·Omni

    ≈12 GB

    - [2026.04.30] Release the preview version of the 8-step inference model SenseNova-U1-8B-MoT-8step-preview. In most cases, the image generation quality of this model closely matches that of the base model (see comparison and existing issues).

    17.6B256K ctx Apache 2.0
    706Apr 2026
    • Same category
    • Larger (17.6B vs 2B)
    • Commercial use OK
    vs
  13. We have developed and released the family Ichigo-llama3s. This family is natively understanding audio and text input.

    8B128K ctx Apache 2.0
    46Sept 2024gguf
    • Same category
    • Larger (8B vs 2B)
    • Commercial use OK
    vs
  14. adventists-ai logo

    Sophon-Realtime-v1

    adventists-ai·Omni

    ? GB

    An open full-duplex voice model family that hears your tone and speaks with one. Adventists AI · Apache-2.0 · Recommended: Tag · Flash · Try it online · GitHub

    ?BApache 2.0
    0Nov 2026
    • Same category
    • Commercial use OK
    • Newer release
    vs
  15. inclusionAI logo

    Duo-1.0-distill

    inclusionAI·Omni

    ? GB

    Duo is a full-duplex multimodal world model for real-time audio-visual interaction. This repository contains the four-step distilled Realization weights for Duo 1.0.

    ?Bminimax-h3-community-license
    0Nov 2026
    • Same category
    • Newer release
    vs
  16. sensenova logo≈12 GB

    SenseNova-U1.5 (Preview) is a natively unified multimodal model built on NEO-unify, featuring new patch encoding and decoding layers.

    17.5B256K ctx Apache 2.0
    166Jul 2026gguf
    • Same category
    • Larger (17.5B vs 2B)
    • Commercial use OK
    vs
  17. mistralai logo

    Voxtral-Small-24B-2507

    mistralai·Omni

    ≈16 GB

    Voxtral Small is an enhancement of Mistral Small 3, incorporating state-of-the-art audio input capabilities while retaining best-in-class text performance. It excels at speech transcription, translation and audio understanding.

    24.3B128K ctx Apache 2.0
    7.5KJul 2025gguf
    • Same category
    • Larger (24.3B vs 2B)
    • Commercial use OK
    vs
  18. Since the research community currently lacks a general-purpose audio captioning model, we fine-tuned Qwen3-Omni-30B-A3B to obtain Qwen3-Omni-30B-A3B-Captioner, which produces detailed, low-hallucination captions for arbitrary audio inputs.

    31.7BA3B64K ctx Apache 2.0
    4.4KSept 2025gguf
    • Same category
    • Larger (31.7B vs 2B)
    • Commercial use OK
    vs
  19. gpt-omni logo

    mini-omni2

    gpt-omni·Omni

    ? GB

    Mini-Omni2 is an omni-interactive model. It can understand image, audio and text inputs and has end-to-end voice conversations with users. Featuring real-time voice output, omni-capable multimodal understanding and flexible interaction ability with interruption mechanism while speaking.

    ?BMIT
    137Oct 2024gguf
    • Same category
    • Commercial use OK
    vs
  20. sensenova logo≈12 GB

    - [2026.04.30] Release the preview version of the 8-step inference model SenseNova-U1-8B-MoT-8step-preview. In most cases, the image generation quality of this model closely matches that of the base model (see comparison and existing issues).

    17.6B256K ctx Apache 2.0
    162Jul 2026
    • Same category
    • Larger (17.6B vs 2B)
    • Commercial use OK
    vs
  21. inclusionAI logo

    Ming-flash-omni-Preview

    inclusionAI·Omni

    ≈63 GB

    Ming-flash-omni Preview, an upgraded version of Ming-Omni, built upon a sparser Mixture-of-Experts (MoE) variant of Ling-Flash-2.0 with 100B total parameters, of which only 6B are active per token.

    104BA7.6B32K ctx MIT
    2.8KOct 2025
    • Same category
    • Larger (104B vs 2B)
    • Commercial use OK
    vs
  22. meituan-longcat logo

    LongCat-Next

    meituan-longcat·Omni

    ≈47 GB

    We develop LongCat-Next, a native multimodal model that processes text, vision, and audio under a single autoregressive objective with minimal inductive bias beyond the language paradigm.

    74.3B128K ctx MIT
    766Mar 2026
    • Same category
    • Larger (74.3B vs 2B)
    • Commercial use OK
    vs
  23. stepfun-ai logo

    Step-Audio-R1.1

    stepfun-ai·Omni

    ≈22 GB

    Step-Audio R1.1 (Realtime) is a major upgrade to Step-Audio-R1, designed for interactive spoken dialogue with both real-time responsiveness and strong reasoning capability.

    33.5B64K ctx Apache 2.0
    948Jan 2026
    • Same category
    • Larger (33.5B vs 2B)
    • Commercial use OK
    vs
  24. Tele-AI logo

    TeleStyleV2

    Tele-AI·Omni

    ? GB

    If you find TeleStyleV2 useful for your research and applications, please cite using this BibTeX:

    ?BApache 2.0
    0Jun 2026
    • Same category
    • Commercial use OK
    vs