Yes — with 5.3 GB to spare
Qwen2.5-Coder 7B at Q4_K_M fits your GeForce RTX 5070 entirely on the GPU at 8K context, at an estimated 95 tokens per second. Past 104K the KV cache pushes it over — quantise the cache to q8_0, or step down a quantisation, to go longer.
Fully on GPU
8K context
Q4_K_M · 4.3 GB
Apache 2.0
Released Nov 2024
The standard local autocomplete model — small enough to keep resident all day, and still the best FIM model under 8B.
The VRAM budget
weights 4.3 GB
Weights 4.3 GB
KV cache @ 8K 0.44 GB
Runtime overhead 0.6 GB
Free 5.3 GB of 10.6 GB
Quantisation ladder
| Quant | Weights | Total @ 8K | Max context | Tok/s | Quality | Fit |
|---|---|---|---|---|---|---|
| F16 | 14.2 GB | 15.2 GB | — | ~6.6 | Reference | 4.6 GB over |
| Q8_0 | 7.5 GB | 8.6 GB | 44K | 54 | −0.1% ppl | Long context |
| Q6_K | 5.8 GB | 6.9 GB | 76K | 70 | −0.4% ppl | Long context |
| Q5_K_M | 5.0 GB | 6.1 GB | 90K | 81 | −0.8% ppl | Long context |
| Q4_K_M | 4.3 GB | 5.3 GB | 104K | 95 | −1.9% ppl | Recommended |
| Q3_K_M | 3.5 GB | 4.5 GB | 119K | 117 | −5.4% ppl | Long context |
| Q2_K | 3.0 GB | 4.0 GB | 128K | 137 | −15% ppl | Long context |
Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it.
How to run it
$ ollama pull qwen2.5-coder:7b $ OLLAMA_CONTEXT_LENGTH=8192 \ ollama run qwen2.5-coder:7b
The default. One binary, a model registry, an OpenAI-compatible port. More on Ollama.
01Download is 4.3 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02Close anything else holding VRAM. A browser with hardware acceleration can sit on 1–2 GB.
03There is room to go to 104K context on this card.