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 plamo-3-translate-8b

70 open-weight models that do the same job as plamo-3-translate-8b (8.5B translation), 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
pfnet logo

plamo-3-translate-8b

pfnet·Translation

≈11 GB

PLaMo 3 Translate 8B is a translation-specialized LLM developed by Preferred Networks (PFN) that delivers world-class translation quality.

8.5B256K ctx plamo-community-license
1.6KOct 2026gguf
--vram
any
--gpus
--quant
  1. Unbabel logo

    wmt22-comet-da

    Unbabel·Translation

    ? GB

    This is a COMET evaluation model: It receives a triplet with (source sentence, translation, reference translation) and returns a score that reflects the quality of the translation compared to both source and reference.

    ?BApache 2.0
    8.7KFeb 2023
    • Same category
    • Commercial use OK
    vs
  2. google-t5 logo

    t5-base

    google-t5·Translation

    ≈1.0 GB

    With T5, we propose reframing all NLP tasks into a unified text-to-text-format where the input and output are always text strings, in contrast to BERT-style models that can only output either a class label or a span of the input.

    223M512 ctx Apache 2.0
    2.8MMar 2022gguf
    • Same category
    • Smaller (223M vs 8.5B)
    • Commercial use OK
    • Widely used
    vs
  3. Unbabel logo

    wmt22-cometkiwi-da

    Unbabel·Translation

    ? GB

    This is a COMET quality estimation model: It receives a source sentence and the respective translation and returns a score that reflects the quality of the translation.

    ?BCC BY-NC-SA 4.0
    4KFeb 2023
    • Same category
    vs
  4. facebook logo

    nllb-200-distilled-600M

    facebook·Translation

    ≈1.8 GB

    This is the model card of NLLB-200's distilled 600M variant.

    600M1K ctx CC BY-NC 4.0
    732.9KJul 2022gguf
    • Same category
    • Smaller (600M vs 8.5B)
    vs
  5. ai4bharat logo

    indictrans2-en-indic-1B

    ai4bharat·Translation

    ≈3.0 GB

    This is the model card of IndicTrans2 En-Indic 1.1B variant.

    1.1BMIT
    49KSept 2023
    • Same category
    • Smaller (1.1B vs 8.5B)
    • Commercial use OK
    vs
  6. ai4bharat logo

    indictrans2-en-indic-dist-200M

    ai4bharat·Translation

    ≈1.1 GB

    This is the model card of IndicTrans2 En-Indic Distilled 200M variant.

    275MMIT
    310.2KSept 2023
    • Same category
    • Smaller (275M vs 8.5B)
    • Commercial use OK
    vs
  7. Helsinki-NLP logo

    opus-mt-fr-en

    Helsinki-NLP·Translation

    ≈0.7 GB

    opus-mt-fr-en is an open-weight translation model with 75.2M parameters from Helsinki-NLP, published on Hugging Face.

    75M512 ctx Apache 2.0
    333.5KMar 2022gguf
    • Same category
    • Smaller (75M vs 8.5B)
    • Commercial use OK
    vs
  8. facebook logo

    nllb-moe-54b

    facebook·Translation

    ≈65 GB

    The NLLB model was presented in No Language Left Behind: Scaling Human-Centered Machine Translation by Marta R. Costa-jussà, James Cross, Onur Çelebi, Maha Elbayad, Kenneth Heafield, Kevin Heffernan, Elahe Kalbassi, Janice Lam, Daniel Licht, Jean Maillard, Anna Sun, Skyler Wang, Guillaume Wenzek…

    54B1K ctx CC BY-NC 4.0
    2.4KMar 2023
    • Same category
    • Larger (54B vs 8.5B)
    vs
  9. LiquidAI logo

    LFM2-350M-ENJP-MT

    LiquidAI·Translation

    ≈1.3 GB

    Based on the LFM2-350M model, this checkpoint has been fine-tuned for near real-time bi-directional Japanese/English translation of short-to-medium inputs.

    354M125K ctx lfm1.0
    1KSept 2025gguf
    • Same category
    • Smaller (354M vs 8.5B)
    vs
  10. CohereLabs logo

    North-Small-Translate-1.0

    CohereLabs·Translation

    ≈260 GB

    North Small Translate is an open weights research release of a sparse Mixture-of-Experts model with 25 billion active parameters and 218 billion total parameters, specialized for high-quality machine translation across 50 languages.

    218BCC BY-NC 4.0
    32Aug 2026
    • Same category
    • Larger (218B vs 8.5B)
    vs
  11. mistralai logo≈5.8 GB

    Voxtral Mini 4B Realtime 2602 is a multilingual, realtime speech-transcription model and among the first open-source solutions to achieve accuracy comparable to offline systems with a delay of <500ms.

    4.4BApache 2.0
    1.5MJan 2026gguf
    • Related category
    • Similar size (4.4B vs 8.5B)
    • Commercial use OK
    • Widely used
    vs
  12. Unbabel logo

    Tower-Plus-9B

    Unbabel·Chat

    ≈8.6 GB

    Tower+ 9B is build on top of Gemma 2 9B. The model goes through the Continuous Pretraining (CPT), Instruction Tuning (IT), Weighted Preference Optimization (WPO). During all stages we include parallel and multilingual data (covering 22 languages).

    9.2B8K ctx CC BY-NC-SA 4.0
    16.9KJun 2025gguf
    • Related category
    • Similar size (9.2B vs 8.5B)
    vs
  13. mistralai logo≈5.8 GB

    Voxtral Mini 4B Realtime Arabic is a streaming speech-to-text model for Arabic dialects and Modern Standard Arabic. It is designed to handle challenging real-world audio and code-switching, where speakers alternate between languages within the same conversation.

    4.4BUnknown licence
    2.2KOct 2026gguf
    • Related category
    • Similar size (4.4B vs 8.5B)
    vs
  14. Qwen logo≈5.8 GB

    The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni.

    2.3BApache 2.0
    1.6MJan 2026gguf
    • Related category
    • Smaller (2.3B vs 8.5B)
    • Commercial use OK
    • Widely used
    vs
  15. Qwen logo≈5.1 GB

    The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni.

    2BApache 2.0
    347.7KJun 2026
    • Related category
    • Smaller (2B vs 8.5B)
    • Commercial use OK
    vs
  16. nvidia logo

    canary-qwen-2.5b

    nvidia·STT

    ≈6.2 GB

    NVIDIA NeMo Canary-Qwen-2.5B is an English speech recognition model that achieves state-of-the art performance on multiple English speech benchmarks.

    2.6BCC BY 4.0
    25KJun 2025gguf
    • Related category
    • Smaller (2.6B vs 8.5B)
    • Commercial use OK
    vs
  17. nyralabs logo≈4.0 GB

    The most accurate verbatim speech recognition you can run in production: controllable, multilingual, and timed to the word.

    1.5Bnyra-health-non-commercial-research
    24.6KJul 2026
    • Related category
    • Smaller (1.5B vs 8.5B)
    vs
  18. facebook logo≈5.7 GB

    SeamlessM4T is our foundational all-in-one Massively Multilingual and Multimodal Machine Translation model delivering high-quality translation for speech and text in nearly 100 languages.

    2.3BCC BY-NC 4.0
    342.9KNov 2023
    • Related category
    • Smaller (2.3B vs 8.5B)
    vs
  19. openai logo

    whisper-large-v3

    openai·STT

    ≈4.0 GB

    Whisper is a state-of-the-art model for automatic speech recognition (ASR) and speech translation, proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI.

    1.5BApache 2.0
    3.2MNov 2023gguf
    • Related category
    • Smaller (1.5B vs 8.5B)
    • Commercial use OK
    • Widely used
    vs
  20. Qwen logo≈2.3 GB

    The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni.

    782MApache 2.0
    535KJun 2026gguf
    • Related category
    • Smaller (782M vs 8.5B)
    • Commercial use OK
    vs
  21. Systran logo≈4.0 GB

    This repository contains the conversion of openai/whisper-large-v3 to the CTranslate2 model format.

    1.6BMIT
    1.7MNov 2023
    • Related category
    • Smaller (1.6B vs 8.5B)
    • Commercial use OK
    • Widely used
    vs
  22. Qwen logo≈2.6 GB

    The Qwen3-ASR family includes Qwen3-ASR-1.7B and Qwen3-ASR-0.6B, which support language identification and ASR for 52 languages and dialects. Both leverage large-scale speech training data and the strong audio understanding capability of their foundation model, Qwen3-Omni.

    938MApache 2.0
    232.4KJan 2026gguf
    • Related category
    • Smaller (938M vs 8.5B)
    • Commercial use OK
    vs
  23. nvidia logo

    canary-1b-v2

    nvidia·STT

    ≈2.7 GB

    Canary-1b-v2 is a powerful 1-billion parameter model built for high-quality speech transcription and translation across 25 European languages.

    979MCC BY 4.0
    228.6KAug 2025gguf
    • Related category
    • Smaller (979M vs 8.5B)
    • Commercial use OK
    vs
  24. openai logo≈2.3 GB

    Whisper is a state-of-the-art model for automatic speech recognition (ASR) and speech translation, proposed in the paper Robust Speech Recognition via Large-Scale Weak Supervision by Alec Radford et al. from OpenAI.

    809MMIT
    4.7MOct 2024gguf
    • Related category
    • Smaller (809M vs 8.5B)
    • Commercial use OK
    • Widely used
    vs