Flywheel-Local-Coder-32B

A 32-billion-parameter coding model in a single Q4_K_M file that runs entirely on your own machine. It takes Qwen2.5-Coder-32B-Instruct and continues its pretraining on the same 66-million-token corpus behind the 14B, drawn from a real working development ecosystem, then packs the merge into one GGUF just under 19 GB. Your prompts and your code never leave your disk. And if you ever care to look, the whole build, corpus to weights, can be retraced hash by hash.

This is the larger sibling of Flywheel-Local-Coder-14B: more capacity, the same local-first stance and the same retraceable chain.

Run it in two commands

hf download zaindanaharper/flywheel-local-coder-32b telos-coder-32b-cpt2019-q4_k_m.gguf --local-dir .
llama-cli -m telos-coder-32b-cpt2019-q4_k_m.gguf -cnv

Prefer Ollama? Download the repo folder so the GGUF and the Modelfile sit together, then:

ollama create flywheel-local-coder-32b -f Modelfile
ollama run flywheel-local-coder-32b

No conversion step, no shards, no Python environment. The usage guide covers chat, deterministic completion, an OpenAI-compatible local endpoint, and how to verify your download against the published checksums.

Specs at a glance

Parameters	32.5B (qwen2 architecture)
Context length	32,768 tokens
Quantization	Q4_K_M, single GGUF file
File size	18.5 GB (19,851,336,480 bytes)
Capabilities	chat, code completion, tool calling
Base model	Qwen2.5-Coder-32B-Instruct
Training	QLoRA continued pretraining, 66.2M tokens across 17,997 files
License	Apache-2.0 (with Qwen attribution)
SHA-256	65e6133fbe4d12579a776047a71bebb98ab86f9e3d343ed821b51dac0ce312f4

Full details in the spec sheet.

What to expect

This is a local-first coding companion at 32B scale: completions, functions, refactors, and tool-calling on your own hardware, with your code staying home. It needs more room than the 14B, roughly 20 GB of memory for the weights plus context, so it is happiest on a 24 GB GPU or a machine with generous RAM; see the spec sheet for hardware guidance.

We publish measurements, not adjectives, and we are precise about which we have. Unlike the 14B, this model does not yet carry benchmark scores. Its only recorded behavioral receipt today is a deterministic generation smoke: at temperature 0 with a fixed seed, reruns are byte-identical. We make no capability-uplift claim over the base model. Benchmarks are pending and will ship with the JSON they came from and the method to re-run them, exactly as the 14B's benchmarks page already does. Until then, the honest statement is: a verified, retraceable continued-pretraining build, not a scored one.

Also in this repo

• telos-coder-32b-cpt2019-q4_k_m.gguf: the runnable merged model (above).
• telos-coder-32b-cpt2019-lora.gguf: the QLoRA adapter (268 MB). Apply the trained delta to your own copy of the base, or requantize from it at another precision. The base weights themselves are never republished here.

The documents

• Usage guide: run it with Ollama, llama.cpp, or as a local API.
• Benchmarks: what is measured so far, and what is still pending.
• Spec sheet: hardware guidance, training details, formats.
• Safety and claims: what this model does and does not claim.
• Model card: the full technical card.
• provenance.json and checksums.sha256: the retraceable chain from corpus to the exact bytes you downloaded.

License and attribution

Apache-2.0. Built on Qwen2.5-Coder-32B-Instruct by the Qwen team; see LICENSE for the attribution notice. The training corpus is proprietary to the author; the shipped weights carry no third-party code beyond the base model, and the base model is never republished on its own.