Skip to content
Computer vision

Tin 0.8B

A language model for programming in Tin

2 min read

Tin 0.8B is a compact language model (LLM (large language model)) with 0.8B parameters, trained to write and fix code in the typed programming language Tin, where agents and tools are built into the syntax. Built on Qwen3.5-0.8B via the loopl-0.8B intermediate checkpoint, the model is designed to work with the Tin compiler in a check-and-fix loop. It helps developers automate code writing and agent integration in applications.

Key features

  • Narrow specialization: trained on Tin syntax and built-in tin_check and tin_run tools.
  • Error fixing: demonstrates high accuracy (up to 88.8%) in fixing bugs based on compiler messages.
  • Lightweight: the 0.8B model size is suitable for local running and further fine-tuning, also available in a 4bit version.
  • Open license: distributed under the Apache-2.0 license, similar to the base Qwen3.5 model.

What you can do

  • Write code in Tin: generate basic programs and functions with support for built-in tools.
  • Fix bugs: feed compiler errors to the model for automated code correction.
  • Test agents: use alongside tin_check and tin_run tools for automated pipelines.

Example

curl -fsSL [https://tin.ad/install.sh](https://tin.ad/install.sh) | bash
tin model pull tin:0.8b
Read moreDetails
Benchmarks (success % on held-out set): Tin knowledge — 50.1%, code writing — 37.5%, bug fixing — 88.8%.
Limitations: about 62% of initial attempts require compiler refinement, may not know new standard library features.
Discuss Telegram 0

Will it run?

Runs in BF16

Fits without quantisation.

  • BF16weights 1.6 GB · ready files2.2 GB
  • FP8 / INT8recommendedweights 0.8 GB1.4 GB
  • 4-bitweights 0.4 GB1.0 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.

What to get for this model

Massed ComputeOpen Massed Compute

Your own GPU: FP8 / INT8 — from RTX 2060 · BF16 — from RTX 2060

Massed Compute hourly prices, November 2026. The discounted price applies with the coupon at checkout.

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.

    # cagataydev/tin-0.8b
    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 cagataydev/tin-0.8b --local-dir ~/models/tin-0.8b
    # The model is in ~/models. How to run it is in the model card (link above).
    Add more LoRAs to tin-0.8b
  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

  1. Qwen3.5-0.8B-Basefamily root
  2. Qwen3.5-0.8B
  3. loopl-0.8b
  4. tin-0.8b

New posts about versions of Qwen3.5-0.8B-Base (quants, finetunes, LoRAs) will show up under the bell.

Quantised versions 12All on Hugging Face
Fine-tunes 12All on Hugging Face
LoRAs and adapters 9All on Hugging Face
Merges 1All on Hugging Face

About the model

cagataydev/tin-0.8b
Author
cagataydev
Category
Computer vision
Released
Posted

On NeyroLearn

Hugging Face

00

Ask AI about this model

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

Comments