Keeping Continue + Qwen 7B within VRAM on Ubuntu
A GitHub issue documents that Continue used ~9.5GB VRAM vs ~6GB for the same model in the Ollama CLI, fixed by lowering contextLength and maxTokens.
Solo2-10 peopleCodingUse existing hardwareSensitivity: Standard
Hardware
| Item | Approx. price (as of 2026-10-16) |
|---|---|
| Any NVIDIA GPU (Linux workstation) | Not stated in source |
Source prices are as stated by the source. Retailer offers: prices checked 2026-10-16, may change. Some links may be affiliate links; we may earn a commission at no extra cost to you. This never changes what we list.
Software stack
- Ubuntu 24.04
- Ollama
- Continue 1.0.18 (VS Code)
Models used
- qwen2.5-coder:7b-instruct
Good at
- Simple config fix restores speed
Weak at
- Default context settings can inflate VRAM and slow completions
Rough total cost
Uses existing hardware.
Getting started
- Run the model in Ollama CLI and note VRAM with nvidia-smi.
- Open Continue config.yaml.
- Under defaultCompletionOptions set contextLength (e.g. 1024-8192) and maxTokens: 500.
- Reload and re-check VRAM.
Source & credit
Issue by @aidinmc — GitHub continuedev/continue #6609
Summarized from the source; quotes are the source author's words. Check the original before buying.