Skip to content
Computer vision

OpenCSG Agentic-27B-V2

a strong agentic fine-tune for automation and coding

2 min read

Built on Qwen/Qwen3.8-27B, OpenCSG Agentic-27B-V2 is trained with LoRA SFT on trajectories collected from real failures that were reproduced and verified in controlled environments. Rather than blindly generating synthetic tasks, this pipeline runs real user errors through live environments and trains the model only on trajectories that pass deterministic verifiers. On held-out benchmarks, it easily outperforms its base model and even beats the teacher model DeepSeek-V4.1-Flash in business automation. It fits on a single 80GB GPU and handles complex production workflows.

Key features

  • Outperforms its teacher: scores 41.67% on AutomationBench, beating the DeepSeek-V4.1-Flash teacher model (33.3%).
  • Trained on real bug fixes: training data isn't generated blindly, but gathered from actual platform failures re-verified inside runtime environments.
  • Dual capability gains: simultaneous improvements in business automation (AutomationBench) and software engineering (SWE-bench Pro).
  • Efficient scale: a 27B dense architecture in BF16 that runs on a single 80GB GPU.

What you can do

  • Deploy a local API service for automation and coding with OpenAI-compatible endpoints.
  • Automate cross-system business workflows and SaaS pipelines (CRMs, email, spreadsheets).
  • Integrate it into dev environments to track down and patch bugs in real codebases.

Example

vllm serve ./Agentic-27B-V2 \
  --served-model-name OpenCSG/Agentic-27B-V2 \
  --tensor-parallel-size 1 \
  --gpu-memory-utilization 0.88 \
  --max-model-len 32768 \
  --enable-auto-tool-choice \
  --tool-call-parser qwen3_xm
Read moreDetails
• Base model: Qwen/Qwen3.8-27B
• Benchmarks: AutomationBench 41.67% (+38.9% rel.), SWE-bench Pro 91.67%
• Hardware: 1 × 80GB GPU (BF16), 262,144 max context tokens
• Teacher: DeepSeek-V4.1-Flash-1017
Examples
Discuss Telegram 0 75

Will it run?

Runs in 4-bit

Full precision does not fit, a quantised version does: quality barely changes down to 8 bits, more noticeably at 4 bits.

  • BF16weights 52 GB · ready files57 GB
  • FP8 / INT8recommendedweights 26 GB29 GB
  • 4-bitweights 15 GB17 GB

Tip: go 8-bit.FP8 / INT8 is within about a percent of full precision in quality, takes half the memory and runs faster on RTX 40/50 and H100. Full precision only pays off for fine-tuning.

Estimate: weights plus working memory. Long videos, high resolutions and long contexts need more.

Run it on Massed Compute

The script downloads the model to the rented machine. How to run it is in the model card.

  1. 1. Rent a machine with a fitting GPU

    Open Massed Compute
  2. 2. Paste the script into the machine's terminal

    Connect over SSH (the Massed Compute panel shows the command) or open a terminal on the machine's desktop.

    HF_TOKEN / CIVITAI_TOKENThe script starts with token lines: paste yours between the quotes. The Hugging Face token is for gated models, the Civitai one for Civitai models.

    # opencsg/Agentic-27B-V2
    mkdir -p ~/models
    curl -LsSf https://astral.sh/uv/install.sh | sh
    source "$HOME/.local/bin/env"
    uv tool install "huggingface_hub[cli]"
    hf download opencsg/Agentic-27B-V2 --local-dir ~/models/Agentic-27B-V2
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to Agentic-27B-V2
  3. 3. Use it

    The model is in ~/models. How to run it is in the model card (link above).

Stop the machine in the panel when you are done: it bills by the minute while it runs.

How to speed it up

Picked for this model type. Choose your hardware above to see what fits you exactly.

  • FlashAttention 44.0.0b40from November 18, 2026new release

    fastest exact attention on B200

    Kernels for Blackwell (beta). Exact attention without quantisation — fine for everything, LLMs included.

    How to enable: pip install flash-attn-4; vLLM and SGLang pick it up themselves.

    Earlier generations (2)
    • FlashAttention 3 — The Hopper (H100/H200) version with FP8.
    • FlashAttention 2 — The standard for RTX 30/40/50 and A100: exact attention without the memory cost of long contexts.
  • torch.compile2.15.0from October 28, 2026new release

    10–30% for free

    Compiles the model for your GPU. Slower first run, faster afterwards.

    How to enable: model = torch.compile(model); in ComfyUI the TorchCompileModel node.

Model family

New posts about versions of Qwen3.8-27B (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 2All on Hugging Face

About the model

opencsg/Agentic-27B-V2
Author
opencsg
Category
Computer vision
Released
Posted

On NeyroLearn

Hugging Face

750

Ask AI about this model

Answered by Gemini 4 · the chat is kept only in this tab

Comments