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 RWKV7-G1k-13.3B-20260930

120 open-weight models that do the same job as RWKV7-G1k-13.3B-20260930 (13.3B 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
RWKV logo≈16 GB

This is an official BlinkDL release of RWKV-7 Goose in Hugging Face Transformers format. RWKV-7 is an attention-free recurrent architecture with a constant-size recurrent state and constant inference work per generated token. Training remains parallelizable.

13.3BApache 2.0
438Sept 2026
--vram
any
--gpus
--quant
  1. ibm-granite logo

    granite-3.3-8b-instruct

    ibm-granite·Chat

    ≈6.6 GB

    Model Summary: Granite-3.3-8B-Instruct is a 8-billion parameter 128K context length language model fine-tuned for improved reasoning and instruction-following capabilities.

    8.2B128K ctx Apache 2.0
    100.7KApr 2025ollama
    • Same category
    • Similar size (8.2B vs 13.3B)
    • Commercial use OK
    vs
  2. LiquidAI logo

    LFM2.5-8B-A1B-Base

    LiquidAI·Chat

    ≈5.9 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.

    8.5BA1B125K ctx lfm1.0
    3.1KMay 2026gguf
    • Same category
    • Similar size (8.5B vs 13.3B)
    vs
  3. nvidia logo≈6.1 GB

    Nemotron-Labs-Diffusion is a tri-mode language model that supports both AR decoding and diffusion-based parallel decoding by simply switching the attention pattern of the same model during inference.

    8.5B4K ctx nvidia-nemotron-open-model-license
    8.4KJan 2026
    • Same category
    • Similar size (8.5B vs 13.3B)
    • Commercial use OK
    vs
  4. Salesforce logo

    Llama-xLAM-2-8b-fc-r

    Salesforce·Chat

    ≈6.3 GB

    Large Action Models (LAMs) are advanced language models designed to enhance decision-making by translating user intentions into executable actions.

    8B128K ctx CC BY-NC 4.0
    137.8KMar 2025gguf
    • Same category
    • Similar size (8B vs 13.3B)
    vs
  5. ibm-granite logo

    granite-4.2-30b

    ibm-granite·Chat

    ≈20 GB

    Granite is a family of open-source large language models developed by IBM, designed for enterprise and research use. Granite models are built to be versatile, safe, and efficient — covering a range of sizes and capabilities from compact edge-deployable models to large-scale reasoning systems.

    29.3B128K ctx Apache 2.0
    37.4KAug 2026ollama
    • Same category
    • Larger (29.3B vs 13.3B)
    • Commercial use OK
    vs
  6. inclusionAI logo

    LLaDA2.0-mini-preview

    inclusionAI·Chat

    ≈10 GB

    LLaDA2.0-mini-preview is a diffusion language model featuring a 16BA1B Mixture-of-Experts (MoE) architecture. As an enhanced, instruction-tuned iteration of the LLaDA series, it is optimized for practical applications.

    16.3BA1.4B8K ctx Apache 2.0
    1.7KOct 2025gguf
    • Same category
    • Similar size (16.3B vs 13.3B)
    • Commercial use OK
    vs
  7. LGAI-EXAONE logo

    EXAONE-3.5-7.8B-Instruct

    LGAI-EXAONE·Chat

    ≈6.1 GB

    We introduce EXAONE 3.5, a collection of instruction-tuned bilingual (English and Korean) generative models ranging from 2.4B to 32B parameters, developed and released by LG AI Research.

    7.8B32K ctx exaone
    383.6KDec 2024ollama
    • Same category
    • Similar size (7.8B vs 13.3B)
    vs
  8. microsoft logo≈9.6 GB

    The Phi-3-Medium-4K-Instruct is a 14B 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.

    14B4K ctx MIT
    23.7KMay 2024ollama
    • Same category
    • Similar size (14B vs 13.3B)
    • Commercial use OK
    vs
  9. openai logo≈14 GB

    gpt-oss-safeguard-120b and gpt-oss-safeguard-20b are safety reasoning models built-upon gpt-oss. With these models, you can classify text content based on safety policies that you provide and perform a suite of foundational safety tasks. These models are intended for safety use cases.

    21.5BA4.8B128K ctx Apache 2.0
    27.2KSept 2025ollama
    • Same category
    • Similar size (21.5B vs 13.3B)
    • Commercial use OK
    vs
  10. openbmb logo

    MathForm-8B

    openbmb·Chat

    ≈6.5 GB

    MathForm-8B is an autoformalization model that translates natural-language mathematical statements into Lean 4. It is released with the paper MathForm: Scaling Mathematical Autoformalization with Knowledge Retrieval and Verification-Guided Refinement.

    8.2B40K ctx Apache 2.0
    2.2KAug 2026gguf
    • Same category
    • Similar size (8.2B vs 13.3B)
    • Commercial use OK
    vs
  11. 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
    19KJun 2025gguf
    • Same category
    • Similar size (9.2B vs 13.3B)
    vs
  12. amazon logo

    ALoDLM-8B

    amazon·Chat

    ≈6.5 GB

    ALoDLM is a family of diffusion language models developed at Amazon, available as ALoDLM-1.7B and ALoDLM-8B and initialized from the corresponding Qwen3 backbones.

    8.2B40K ctx CC BY-NC 4.0
    1.5KOct 2026gguf
    • Same category
    • Similar size (8.2B vs 13.3B)
    • Newer release
    vs
  13. gnani logo≈20 GB

    gnani-evon-v3.3 is a hybrid Mamba architecture 30B MOE A3B model trained end-to-end for best-in-class Indic capability — continued pretraining, supervised fine-tuning, and reinforcement learning — delivering deep comprehension, extractive QA, and reasoning across Hindi, Bengali, Telugu, Tamil…

    31.7BA3B128K ctx Apache 2.0
    71.9KAug 2026gguf
    • Same category
    • Larger (31.7B vs 13.3B)
    • Commercial use OK
    vs
  14. Qwen logo

    Qwen3-32B

    Qwen·Chat

    ≈22 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.

    32.8B40K ctx Apache 2.0
    3.4MApr 2025ollama
    • Same category
    • Larger (32.8B vs 13.3B)
    • Commercial use OK
    • Widely used
    vs
  15. Unbabel logo

    M-Prometheus-14B

    Unbabel·Chat

    ≈11 GB

    M-Prometheus is a suite of open LLM judges that can natively evaluate multilingual outputs. They were trained on 480k instances of multilingual direct assessment and pairwise comparison data with long-form feedback. They can be prompted in the same way as Prometheus-2.

    14.8B32K ctx Custom licence
    2.9KApr 2025gguf
    • Same category
    • Similar size (14.8B vs 13.3B)
    vs
  16. Almawave logo

    Velvet-14B

    Almawave·Chat

    ≈10 GB

    Velvet is an Italian family of large language models, developed from scratch, featuring a dense architecture. This model was trained on the HPC Leonardo infrastructure hosted by CINECA, utilizing public data that underwent extensive curation.

    14.1BApache 2.0
    2.8KJan 2025gguf
    • Same category
    • Similar size (14.1B vs 13.3B)
    • Commercial use OK
    vs
  17. deepseek-ai logo

    DeepSeek-V2-Lite-Chat

    deepseek-ai·Chat

    ≈12 GB

    Last week, the release and buzz around DeepSeek-V2 have ignited widespread interest in MLA (Multi-head Latent Attention)! Many in the community suggested open-sourcing a smaller MoE model for in-depth research. And now DeepSeek-V2-Lite comes out:

    15.7BA2.7B160K ctx DeepSeek License
    55.1KMay 2024gguf
    • Same category
    • Similar size (15.7B vs 13.3B)
    vs
  18. Qwen logo≈19 GB

    We introduce the updated version of the Qwen3-30B-A3B non-thinking mode, named Qwen3-30B-A3B-Instruct-2507, featuring the following key enhancements:

    30.5BA3B256K ctx Apache 2.0
    802.5KJul 2025gguf
    • Same category
    • Larger (30.5B vs 13.3B)
    • Commercial use OK
    vs
  19. tencent logo

    ContextPilot-E4B

    tencent·Chat

    ≈5.9 GB

    ContextPilot-E4B is the Gemma4-E4B checkpoint of ContextPilot, a proactive context-management framework for long-horizon language-model agents. It teaches agents to plan, maintain long-term memory, and offload less useful context while they continue reasoning and using tools.

    7.9B128K ctx Custom licence
    2.1KAug 2026gguf
    • Same category
    • Similar size (7.9B vs 13.3B)
    vs
  20. zai-org logo

    GLM-4-9B-0414

    zai-org·Chat

    ≈6.4 GB

    The GLM family welcomes new members, the GLM-4-32B-0414 series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series. It also supports very user-friendly local deployment features.

    9.4B32K ctx MIT
    22.2KApr 2025ollama
    • Same category
    • Similar size (9.4B vs 13.3B)
    • Commercial use OK
    vs
  21. Qwen logo≈10 GB

    Qwen1.5-MoE is a transformer-based MoE decoder-only language model pretrained on a large amount of data.

    14.3BA2.7B8K ctx Qwen License
    40.9KFeb 2024gguf
    • Same category
    • Similar size (14.3B vs 13.3B)
    vs
  22. microsoft logo

    Sentiment-Analysis-7B

    microsoft·Chat

    ≈5.8 GB

    Sentiment-Analysis-7B is a 7B-parameter autoregressive language model fine-tuned for sentiment classification of user messages directed at a service, product, or agent (the "Target").

    7.2B32K ctx MIT
    3KOct 2026gguf
    • Same category
    • Similar size (7.2B vs 13.3B)
    • Commercial use OK
    • Newer release
    vs
  23. NVIDIA-Nemotron-Nano-9B-v2-Japanese is a large language model (LLM) trained from scratch by NVIDIA and designed as a unified model for both reasoning and non-reasoning tasks, specifically optimized for the Japanese language.

    8.9B128K ctx nvidia-nemotron-open-model-license
    3.2KFeb 2026gguf
    • Same category
    • Similar size (8.9B vs 13.3B)
    • Commercial use OK
    vs
  24. NousResearch logo

    Arete-27B

    NousResearch·Chat

    ≈19 GB

    We post-trained over Alibaba's Qwen3.8-27B to reinforce western values on a Chinese-trained model. We call this new model Arete-27B.

    27.4B256K ctx Apache 2.0
    7.5KNov 2026gguf
    • Same category
    • Larger (27.4B vs 13.3B)
    • Commercial use OK
    • Newer release
    vs