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 granite-4.1-3b-llmon-preview

120 open-weight models that do the same job as granite-4.1-3b-llmon-preview (3.7B chat), 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
ibm-granite logo≈3.3 GB

Model Summary: Granite-4.1-3B-LLMON-preview is a 3B parameter instruct model fine-tuned from Granite-4.1-3B-base to consume prompts in LLMON (LLM Object Notation) instead of a conventional chat template.

3.7B128K ctx Apache 2.0
2.2KOct 2026gguf
--vram
any
--gpus
--quant
  1. Nanbeige logo

    Nanbeige4.1-3B

    Nanbeige·Chat

    ≈3.3 GB

    Nanbeige4.1-3B is built upon Nanbeige4-3B-Base and represents an enhanced iteration of our previous reasoning model, Nanbeige4-3B-Thinking-2511, achieved through further post-training optimization with supervised fine-tuning (SFT) and reinforcement learning (RL).

    3.9B256K ctx Apache 2.0
    10.3KFeb 2026gguf
    • Same category
    • Similar size (3.9B vs 3.7B)
    • Commercial use OK
    vs
  2. microsoft logo

    Phi-tiny-MoE-instruct

    microsoft·Chat

    ≈3.0 GB

    Phi-tiny-MoE is a lightweight Mixture of Experts (MoE) model with 3.8B total parameters and 1.1B activated parameters. It is compressed and distilled from the base model shared by Phi-3.5-MoE and GRIN-MoE using the SlimMoE approach, then post-trained via supervised fine-tuning and direct preference…

    3.8BA1.3B4K ctx MIT
    30.3KJun 2025gguf
    • Same category
    • Similar size (3.8B vs 3.7B)
    • Commercial use OK
    vs
  3. ibm-granite logo

    granite-4.0-micro

    ibm-granite·Chat

    ≈3.1 GB

    📣 Update [10-07-2025]: Added a default system prompt to the chat template to guide the model towards more professional, accurate, and safe responses.

    3.4B128K ctx Apache 2.0
    22.2KSept 2025gguf
    • Same category
    • Similar size (3.4B vs 3.7B)
    • Commercial use OK
    vs
  4. IFM logo≈4.6 GB

    K2-Horizon-3.7B is the small dense member of the K2-Horizon family: a 3.7B-core decoder-only model with a 512K context window.

    5.1B512K ctx Apache 2.0
    15.2KSept 2026gguf
    • Same category
    • Similar size (5.1B vs 3.7B)
    • Commercial use OK
    vs
  5. microsoft logo

    Phi-3.5-mini-instruct

    microsoft·Chat

    ≈5.8 GB

    Phi-3.5-mini is a lightweight, state-of-the-art open model built upon datasets used for Phi-3 - synthetic data and filtered publicly available websites - with a focus on very high-quality, reasoning dense data. The model belongs to the Phi-3 model family and supports 128K token context length.

    3.8B128K ctx MIT
    340.3KAug 2024gguf
    • Same category
    • Similar size (3.8B vs 3.7B)
    • Commercial use OK
    vs
  6. TokenRhythm logo

    NeoHorse-1-4B

    TokenRhythm·Chat

    ≈4.0 GB

    NeoHorse-1-4B is a 4B causal language model and an initial prototype on the path toward recursive self-improvement (RSI). It is post-trained from Qwen3.5-4B for text-based agent harnesses, tool use, coding, and instruction following.

    4.2B256K ctx Apache 2.0
    2.8KSept 2026gguf
    • Same category
    • Similar size (4.2B vs 3.7B)
    • Commercial use OK
    vs
  7. HuggingFaceTB logo

    SmolLM3-3B-Base

    HuggingFaceTB·Chat

    ≈2.9 GB

    SmolLM3 is a 3B parameter language model designed to push the boundaries of small models. It supports 6 languages, advanced reasoning and long context. SmolLM3 is a fully open model that offers strong performance at the 3B–4B scale.

    3.1B64K ctx Apache 2.0
    88.2KJun 2025gguf
    • Same category
    • Similar size (3.1B vs 3.7B)
    • Commercial use OK
    vs
  8. CohereLabs logo

    tiny-aya-earth

    CohereLabs·Chat

    ≈3.1 GB

    Cohere Labs Tiny Aya is an open weights research release of a pretrained 3.35 billion parameter model optimized for efficient, strong, and balanced multilingual representation across 70+ languages, including many lower-resourced ones.

    3.3BCC BY-NC 4.0
    6KFeb 2026gguf
    • Same category
    • Similar size (3.3B vs 3.7B)
    vs
  9. microsoft logo

    UserRepGen-4B

    microsoft·Chat

    ≈4.2 GB

    UserRepGen-4B is a fine-tuned Qwen3.5-4B language model that generates a structured, natural-language user representation from a multi-turn user-assistant conversation.

    4.5B256K ctx MIT
    2.5KOct 2026gguf
    • Same category
    • Similar size (4.5B vs 3.7B)
    • Commercial use OK
    • Newer release
    vs
  10. meta-llama logo

    Llama-3.2-3B

    meta-llama·Chat

    ≈3.3 GB

    The Llama 3.2 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction-tuned generative models in 1B and 3B sizes (text in/text out).

    3.2B128K ctx Llama 3.2 Community License
    313.5KSept 2024gguf
    • Same category
    • Similar size (3.2B vs 3.7B)
    vs
  11. openbmb logo

    MiniCPM5-2B-Base

    openbmb·Chat

    ≈2.3 GB

    We are releasing MiniCPM5-2B, the second model in the MiniCPM5 series, following MiniCPM5-1B. It is a dense 2B Transformer that scales up the same training recipe, built for on-device, local deployment, and resource-constrained scenarios, reaching 2B-class open-source SOTA.

    2.5B512K ctx Apache 2.0
    12.8KAug 2026gguf
    • Same category
    • Similar size (2.5B vs 3.7B)
    • Commercial use OK
    vs
  12. janhq logo

    Jan-code-4b

    janhq·Chat

    ≈4.2 GB

    Jan-Code-4B is a code-tuned model built on top of Jan-v3-4B-base-instruct. It’s designed to be a practical coding model you can run locally and iterate on quickly—useful for everyday code tasks and as a lightweight “worker” model in agentic workflows.

    4.4B256K ctx Apache 2.0
    8KMar 2026gguf
    • Same category
    • Similar size (4.4B vs 3.7B)
    • Commercial use OK
    vs
  13. LiquidAI logo

    LFM2.5-2.6B-Base

    LiquidAI·Chat

    ≈2.6 GB

    LFM2.5 is a new family of hybrid models designed for on-device deployment. It builds on the LFM2 architecture with extended pre-training and reinforcement learning.

    2.7B125K ctx lfm1.0
    7.4KAug 2026gguf
    • Same category
    • Similar size (2.7B vs 3.7B)
    vs
  14. ibm-granite logo≈3.0 GB

    Model Summary: Granite-3.1-3B-A800M-Instruct is a 3B parameter long-context instruct model finetuned from Granite-3.1-3B-A800M-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets tailored for solving long context problems.

    3.3BA800M128K ctx Apache 2.0
    114.9KDec 2024gguf
    • Same category
    • Similar size (3.3B vs 3.7B)
    • Commercial use OK
    vs
  15. Qwen logo

    Qwen3-1.7B

    Qwen·Chat

    ≈2.6 GB

    Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.

    2B40K ctx Apache 2.0
    3MApr 2025ollama
    • Same category
    • Similar size (2B vs 3.7B)
    • Commercial use OK
    • Widely used
    vs
  16. ibm-granite logo

    granite-swash-3b-a600m

    ibm-granite·Chat

    ≈2.5 GB

    Model Summary Granite-SWASH-3B-a600M is a 3B parameter decoder-only sparse language model with per-layer Sliding Window Attention (SWA) and learnable attention sinks (LSE-based).

    3BA600M8K ctx Apache 2.0
    3KJul 2026gguf
    • Same category
    • Similar size (3B vs 3.7B)
    • Commercial use OK
    vs
  17. microsoft logo

    Phi-3-mini-4k-instruct

    microsoft·Chat

    ≈4.3 GB

    The Phi-3-Mini-4K-Instruct is a 3.8B parameters, lightweight, state-of-the-art open model trained with the Phi-3 datasets that includes both synthetic data and the filtered publicly available websites data with a focus on high-quality and reasoning dense properties.

    3.8B4K ctx MIT
    310.3KApr 2024ollama
    • Same category
    • Similar size (3.8B vs 3.7B)
    • Commercial use OK
    vs
  18. WeiboAI logo

    VibeThinker-3B

    WeiboAI·Chat

    ≈2.6 GB

    2.For harder math reasoning, try AMOBench , a problem set harder than the International Mathematical Olympiad (IMO), with included standard answers. Use it to evaluate VibeThinker against other SOTA models. Note: due to extreme difficulty, set max tokens to 60K–100K.

    3.1B128K ctx MIT
    2.6KJun 2026gguf
    • Same category
    • Similar size (3.1B vs 3.7B)
    • Commercial use OK
    vs
  19. CohereLabs logo

    tiny-aya-base

    CohereLabs·Chat

    ≈3.1 GB

    Cohere Labs Tiny Aya is an open weights research release of a pretrained 3.35 billion parameter model optimized for efficient, strong, and balanced multilingual representation across 70+ languages, including many lower-resourced ones.

    3.3BCC BY-NC 4.0
    2.9KFeb 2026gguf
    • Same category
    • Similar size (3.3B vs 3.7B)
    vs
  20. ibm-granite logo≈3.6 GB

    Model Summary: Granite Switch 4.1 3B Preview is a modular LLM built on IBM Granite 4.1 3B with embedded adapters from the Granite Libraries collection.

    4.1B128K ctx Apache 2.0
    1.8KMay 2026gguf
    • Same category
    • Similar size (4.1B vs 3.7B)
    • Commercial use OK
    vs
  21. InternScience logo

    Agents-A1-4B

    InternScience·Chat

    ≈4.2 GB

    Agents‑A1 is a long-horizon agentic model from InternScience, built to scale heterogeneous agentic abilities across multiple domains including Long‑horizon Search, Engineering, Scientific Research, Instruction Following, and Tool-calling.

    4.5B256K ctx Apache 2.0
    2.3KJul 2026gguf
    • Same category
    • Similar size (4.5B vs 3.7B)
    • Commercial use OK
    vs
  22. microsoft logo≈5.8 GB

    The Phi-3-Mini-128K-Instruct is a 3.8 billion-parameter, lightweight, state-of-the-art open model trained using the Phi-3 datasets. This dataset includes both synthetic data and filtered publicly available website data, with an emphasis on high-quality and reasoning-dense properties.

    3.8B128K ctx MIT
    240.5KApr 2024gguf
    • Same category
    • Similar size (3.8B vs 3.7B)
    • Commercial use OK
    vs
  23. openbmb logo≈2.5 GB

    This is the RL initialization checkpoint used in the blog JustRL II: Scaling Small LLMs to 128K Reasoning with a Critic (中文版).

    2.8B64K ctx Unknown licence
    2.2KSept 2026
    • Same category
    • Similar size (2.8B vs 3.7B)
    vs
  24. BAAI logo

    AREX-Turbo

    BAAI·Chat

    ≈4.2 GB

    AREX is a family of deep research agents developed by the Beijing Academy of Artificial Intelligence (BAAI). It is designed for long-horizon tasks in which an agent must search across sources, assemble candidate answers, verify multiple constraints, and revise its research plan when the available…

    4.5B256K ctx Apache 2.0
    1.9KJul 2026gguf
    • Same category
    • Similar size (4.5B vs 3.7B)
    • Commercial use OK
    vs