Code Llama 7B Instruct
Meta
Now superseded by Qwen2.5-Coder, but still the reference point for a lot of editor integrations.
Good fit
- Weights
- 6.6 GB
- KV cache
- 4.0 GB
- Overhead
- 570 MB
- 11 GB of 15 GB usable VRAM.
- KV cache at 8,192 tokens is 4.0 GB — shorten the context to reclaim memory.
- Chosen as the best quality that still fits a 8,192-token context (11 GB at that length).
- Ubiquitous editor integration
- Fill-in-the-middle
- Clearly behind modern code models
- Full multi-head attention: heavy KV cache
codellama:7bRunning Code Llama 7B Instruct on GeForce RTX 4080 Super
| Quantization | Quality | Weights | KV cache | Total | ~tok/s | Fit |
|---|---|---|---|---|---|---|
| F16 | lossless | 12 GB | 4.0 GB | 17 GB | 42 | Runs with CPU offload |
| Q8_0Pick | near-lossless | 6.6 GB | 4.0 GB | 11 GB | 80 | Good fit |
| Q6_K | near-lossless | 5.1 GB | 4.0 GB | 9.7 GB | 104 | Good fit |
| Q5_K_M | high | 4.4 GB | 4.0 GB | 9.0 GB | 120 | Excellent fit |
| Q4_K_M | balanced | 3.8 GB | 4.0 GB | 8.3 GB | 141 | Excellent fit |
| Q3_K_M | degraded | 3.0 GB | 4.0 GB | 7.6 GB | 174 | Excellent fit |
KV cache is sized at 8,192 tokens. Longer contexts cost proportionally more — the recommendation above reserves room for a working context.
You can fine-tune this here
- Base weights
- 3.5 GB
- Optimizer
- 596 MB
- Activations
- 820 MB
- Peak
- 6.4 GB
- 6.4 GB peak against 15 GB usable — room to raise batch size or sequence length.
- Base weights
- 12 GB
- Optimizer
- 596 MB
- Activations
- 820 MB
- Peak
- 15 GB
- Needs 15 GB — switch to QLoRA to cut the weight footprint.
Where this model runs
VRAM 32 GB · F16 · 17 GB
VRAM 24 GB · Q8_0 · 11 GB
VRAM 16 GB · Q8_0 · 11 GB
VRAM 16 GB · Q8_0 · 11 GB
VRAM 24 GB · Q8_0 · 11 GB
VRAM 12 GB · Q4_K_M · 8.3 GB
VRAM 12 GB · Q4_K_M · 8.3 GB
VRAM 16 GB · Q8_0 · 11 GB
VRAM 48 GB · F16 · 17 GB
VRAM 80 GB · F16 · 17 GB
Unified 128 GB · F16 · 17 GB
Unified 48 GB · F16 · 17 GB
Unified 24 GB · Q6_K · 9.7 GB
Unified 192 GB · F16 · 17 GB
Unified 16 GB · Q5_K_M · 9.0 GB
VRAM 24 GB · Q8_0 · 11 GB
VRAM 16 GB · Q8_0 · 11 GB
VRAM 0 MB
VRAM 0 MB
Qwen2.5 7B Instruct
7.6B · Apache 2.0
The default starting point for local work on 8–12 GB cards. Strong instruction following and reliable tool-call formatting for its size.
Qwen2.5 Coder 7B Instruct
7.6B · Apache 2.0
The practical local copilot. Supports fill-in-the-middle, so it works as an inline completion model rather than only a chat assistant.
Qwen3 8B
8.2B · Apache 2.0
Switchable thinking mode: the same weights answer directly or reason step by step depending on the prompt. Long context for its size.
Llama 3.1 8B Instruct
8B · Llama 3.1 Community License
The most widely supported open model there is. If a tool, adapter or tutorial exists, it was written for this one first.
Catalogue figures come from each model’s published card. ModelLM has not independently measured them.