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 Cola2

120 open-weight models that do the same job as Cola2 (8.8B 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
ByteDance-Seed logo

Cola2

ByteDance-Seed·Chat

≈6.8 GB

Cola2 is a continuous-latent language model that co-adapts semantic compression and discrete text generation. It combines a dynamic-compression VAE, a block-causal Diffusion Transformer (DiT), and a span-causal token decoder with latent feedback.

8.8B32K ctx Unknown licence
1.3KNov 2026
--vram
any
--gpus
--quant
  1. zgcagi logo

    ZGCM-1-7B

    zgcagi·Chat

    ≈5.9 GB

    ZGCM-1 is a 7.39B-parameter dense language model trained from scratch, built for mathematical reasoning and tool-assisted search. It combines deliberate internal thinking with active information gathering, supporting 256K-token context and both thinking and direct-response modes in a single model.

    7.4B256K ctx MIT
    957Sept 2026gguf
    • Same category
    • Similar size (7.4B vs 8.8B)
    • Commercial use OK
    vs
  2. openbmb logo

    AgentCPM-Report

    openbmb·Chat

    ≈5.6 GB

    This repository contains AgentCPM-Report, an 8B-parameter deep research agent introduced in the paper AgentCPM-Report: Interleaving Drafting and Deepening for Open-Ended Deep Research.

    8.2B64K ctx Apache 2.0
    2.1KJan 2026gguf
    • Same category
    • Similar size (8.2B vs 8.8B)
    • Commercial use OK
    vs
  3. NousResearch logo

    Meta-Llama-3.1-8B-Instruct

    NousResearch·Chat

    ≈6.3 GB

    The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models in 8B, 70B and 405B sizes (text in/text out).

    8B128K ctx Llama 3.1 Community License
    137.4KJul 2024gguf
    • Same category
    • Similar size (8B vs 8.8B)
    vs
  4. Qwen logo

    Qwen2-7B

    Qwen·Chat

    ≈5.5 GB

    Qwen2 is the new series of Qwen large language models. For Qwen2, we release a number of base language models and instruction-tuned language models ranging from 0.5 to 72 billion parameters, including a Mixture-of-Experts model. This repo contains the 7B Qwen2 base language model.

    7.6B128K ctx Apache 2.0
    294.9KJun 2024gguf
    • Same category
    • Similar size (7.6B vs 8.8B)
    • Commercial use OK
    vs
  5. Qwen logo≈5.5 GB

    Qwen2 is the new series of Qwen large language models. For Qwen2, we release a number of base language models and instruction-tuned language models ranging from 0.5 to 72 billion parameters, including a Mixture-of-Experts model. This repo contains the instruction-tuned 7B Qwen2 model.

    7.6B32K ctx Apache 2.0
    311.9KJun 2024ollama
    • Same category
    • Similar size (7.6B vs 8.8B)
    • Commercial use OK
    vs
  6. allenai logo≈8.7 GB

    We expand on our Olmo model series by introducing Olmo Hybrid, a new 7B hybrid RNN model in the Olmo family. Olmo Hybrid dramatically outperforms Olmo 3 in final performance, consistently showing roughly 2x data efficiency on core evals over the course of our pretraining run.

    7.4B32K ctx Apache 2.0
    3KFeb 2026
    • Same category
    • Similar size (7.4B vs 8.8B)
    • Commercial use OK
    vs
  7. google logo

    gemma-2-9b-it

    google·Chat

    ≈8.6 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.

    9.2B8K ctx Gemma Terms of Use
    108.5KJun 2024ollama
    • Same category
    • Similar size (9.2B vs 8.8B)
    vs
  8. ByteDance-Seed logo

    academic-ds-9B

    ByteDance-Seed·Chat

    ≈7.5 GB

    This is a 9B model whose architecture is deepseek-v3, trained from scratch using 350B+ tokens from fully open-source, English-only datasets. It is designed for development and debugging purposes within the open-source community.

    9.4BA1.6B8K ctx Apache 2.0
    12.4KApr 2025gguf
    • Same category
    • Similar size (9.4B vs 8.8B)
    • Commercial use OK
    vs
  9. openbmb logo

    MiniCPM4-8B

    openbmb·Chat

    ≈5.6 GB

    MiniCPM4 series are highly efficient large language models (LLMs) designed explicitly for end-side devices, which achieves this efficiency through systematic innovation in four key dimensions: model architecture, training data, training algorithms, and inference systems.

    8.2B32K ctx Apache 2.0
    8KJun 2025gguf
    • Same category
    • Similar size (8.2B vs 8.8B)
    • Commercial use OK
    vs
  10. openbmb logo

    MiniCPM4-MCP

    openbmb·Chat

    ≈5.6 GB

    MiniCPM4 series are highly efficient large language models (LLMs) designed explicitly for end-side devices, which achieves this efficiency through systematic innovation in four key dimensions: model architecture, training data, training algorithms, and inference systems.

    8.2B32K ctx Apache 2.0
    8.1KJun 2025gguf
    • Same category
    • Similar size (8.2B vs 8.8B)
    • Commercial use OK
    vs
  11. stepfun-ai logo

    PaCoRe-8B

    stepfun-ai·Chat

    ≈6.5 GB

    We introduce PaCoRe (Parallel Coordinated Reasoning), a framework that shifts the driver of inference from sequential depth to coordinated parallel breadth, breaking the model context limitation and massively scaling test time compute: Think in Parallel: PaCoRe launches massive parallel exploration…

    8.2B320K ctx MIT
    2KDec 2025gguf
    • Same category
    • Similar size (8.2B vs 8.8B)
    • Commercial use OK
    vs
  12. ibm-granite logo

    granite-3.1-8b-instruct

    ibm-granite·Chat

    ≈6.6 GB

    Model Summary: Granite-3.1-8B-Instruct is a 8B parameter long-context instruct model finetuned from Granite-3.1-8B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets tailored for solving long context problems.

    8.2B128K ctx Apache 2.0
    31.3KDec 2024gguf
    • Same category
    • Similar size (8.2B vs 8.8B)
    • Commercial use OK
    vs
  13. ibm-granite logo≈7.5 GB

    Model Summary: Granite Switch 4.1 8B Preview is a modular LLM built on IBM Granite 4.1 8B with embedded adapters from the Granite Libraries collection.

    9.6B128K ctx Apache 2.0
    705May 2026
    • Same category
    • Similar size (9.6B vs 8.8B)
    • Commercial use OK
    vs
  14. K-intelligence logo

    Midm-2.0-Base-Instruct

    K-intelligence·Chat

    ≈8.8 GB

    Mi:dm 2.0 is a "Korea-centric AI" model developed using KT's proprietary technology. The term "Korea-centric AI" refers to a model that deeply internalizes the unique values, cognitive frameworks, and commonsense reasoning inherent to Korean society.

    11.5B32K ctx MIT
    40.3KJul 2025gguf
    • Same category
    • Similar size (11.5B vs 8.8B)
    • Commercial use OK
    vs
  15. nvidia logo≈6.8 GB

    NVIDIA-Nemotron-Nano-9B-v2-Base is a large language model (LLM) trained from scratch by NVIDIA that is designed as a completion model for a given piece of text. It uses a hybrid model architecture that consists primarily of Mamba-2 and MLP layers combined with just four Attention layers.

    8.9BNVIDIA Open Model License
    2.7KAug 2025gguf
    • Same category
    • Similar size (8.9B vs 8.8B)
    • Commercial use OK
    vs
  16. XiaomiMiMo logo

    MiMo-7B-Base

    XiaomiMiMo·Chat

    ≈6.3 GB

    Currently, most successful RL works, including open-source research, rely on relatively large base models, e.g., 32B models, particularly for enhancing code reasoning capabilities.

    7.8B32K ctx MIT
    14.9KApr 2025
    • Same category
    • Similar size (7.8B vs 8.8B)
    • Commercial use OK
    vs
  17. google logo≈4.6 GB

    T5Gemma is a family of lightweight yet powerful encoder-decoder research models from Google. These models are created by adapting pretrained decoder-only models into a encoder-decoder.

    5.6B8K ctx Gemma Terms of Use
    197.7KJun 2025
    • Same category
    • Similar size (5.6B vs 8.8B)
    vs
  18. Qwen logo≈4.0 GB

    We introduce the updated version of the Qwen3-4B non-thinking mode, named Qwen3-4B-Instruct-2507, featuring the following key enhancements:

    4B256K ctx Apache 2.0
    2.7MAug 2025gguf
    • Same category
    • Smaller (4B vs 8.8B)
    • Commercial use OK
    • Widely used
    vs
  19. mistralai logo≈5.8 GB

    The Mistral-7B-Instruct-v0.2 Large Language Model (LLM) is an instruct fine-tuned version of the Mistral-7B-v0.2.

    7.2B32K ctx Apache 2.0
    1.2MDec 2023ollama
    • Same category
    • Similar size (7.2B vs 8.8B)
    • Commercial use OK
    vs
  20. openai logo

    gpt-oss-20b

    openai·Chat

    ≈13 GB

    Welcome to the gpt-oss series, OpenAI’s open-weight models designed for powerful reasoning, agentic tasks, and versatile developer use cases.

    20.9BA4.2B128K ctx Apache 2.0
    5.6MAug 2025ollama
    • Same category
    • Larger (20.9B vs 8.8B)
    • Commercial use OK
    • Widely used
    vs
  21. tokyotech-llm logo≈6.3 GB

    Llama 3.1 Swallow is a series of large language models (8B, 70B) that were built by continual pre-training on the Meta Llama 3.1 models. Llama 3.1 Swallow enhanced the Japanese language capabilities of the original Llama 3.1 while retaining the English language capabilities.

    8B128K ctx Llama 3.3 Community License
    7.4KJun 2025gguf
    • Same category
    • Similar size (8B vs 8.8B)
    vs
  22. 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 8.8B)
    • Commercial use OK
    vs
  23. microsoft logo

    UserLM-8b

    microsoft·Chat

    ≈6.3 GB

    Unlike typical LLMs that are trained to play the role of the "assistant" in conversation, we trained UserLM-8b to simulate the “user” role in conversation (by training it to predict user turns in a large corpus of conversations called WildChat).

    8B8K ctx MIT
    2.7KSept 2025gguf
    • Same category
    • Similar size (8B vs 8.8B)
    • Commercial use OK
    vs
  24. allenai logo≈8.8 GB

    We introduce Olmo 3, a new family of 7B and 32B models both Instruct and Think variants. Long chain-of-thought thinking improves reasoning tasks like math and coding.

    7.3B64K ctx Apache 2.0
    3.9KNov 2025gguf
    • Same category
    • Similar size (7.3B vs 8.8B)
    • Commercial use OK
    vs