id: qwen3-coder-30b-a3b-instruct
name: Qwen3 Coder 30B A3B Instruct
type: base-model
task:
  - 1c
  - code
  - agentic-coding
  - repository-analysis
  - tool-use
language:
  - ru
  - en
source: huggingface
upstream_id: Qwen/Qwen3-Coder-30B-A3B-Instruct
upstream_url: https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct
license: apache-2.0
status: staging
created_at: 2026-07-04
storage_path: /models/base/qwen3-coder-30b-a3b-instruct
format: safetensors
quantization: none
context_length: 262144
deployment_context_length: 8192
vram_required_gb: null
parameters: 30.5B total / 3.3B active
base_model: null
adapter_for: null
datasets: []
eval_suites:
  - plugins/1c/evals/smoke.yaml
deployment:
  target: docker-gpu.cin.su
  runtime: training
  served_model_name: qwen3-coder-30b-a3b-instruct
notes: "Training base for the current local 1C/code route. Use this HF checkpoint for LoRA fine-tuning, then convert or merge adapters for the Q6 GGUF deployment."