Meta's Llama-2-based code models. Still popular on Ollama for completion.
coding · sizes: 7B, 13B, 34B · Llama 2 Community
| machine | best fit | verdict | speed |
|---|---|---|---|
| Apple M1 · 8GB | 34B Q3_K_M | ❌ Won't fit | — |
| Apple M1 Pro · 16GB | 13B Q4_K_M | 🐢 CPU-only (slow) | 2–3 tok/s |
| Apple M2 · 16GB | 13B Q4_K_M | 🐢 CPU-only (slow) | 0.9–1 tok/s |
| Apple M3 Pro · 18GB | 7B Q5_K_M | ✅ Runs (tight) | 16–25 tok/s |
| Apple M2 Pro · 32GB | 34B Q3_K_M | ✅ Runs (tight) | 6–11 tok/s |
| Apple M4 Pro · 48GB | 34B Q5_K_M | ✅ Runs comfortably | 6–10 tok/s |
| Apple M3 Max · 36GB | 34B Q3_K_M | ✅ Runs (tight) | 11–18 tok/s |
| Apple M2 Max · 64GB | 34B Q8_0 | ✅ Runs (tight) | 5–8 tok/s |
| Apple M4 Max · 128GB | 34B Q8_0 | ✅ Runs comfortably | 7–11 tok/s |
| Apple M2 Ultra · 192GB | 34B Q8_0 | ✅ Runs comfortably | 9–15 tok/s |
| NVIDIA GeForce RTX 3060 · 12GB VRAM | 7B Q5_K_M | ✅ Runs (tight) | 30–50 tok/s |
| NVIDIA GeForce RTX 4070 · 12GB VRAM | 7B Q5_K_M | ✅ Runs (tight) | 40–70 tok/s |
| NVIDIA GeForce RTX 4090 · 24GB VRAM | 34B Q4_K_M | ✅ Runs (tight) | 20–35 tok/s |
| NVIDIA GeForce RTX 5080 · 16GB VRAM | 13B Q4_K_M | ✅ Runs (tight) | 50–80 tok/s |
| NVIDIA GeForce RTX 5090 · 32GB VRAM | 34B Q5_K_M | ✅ Runs (tight) | 30–50 tok/s |
| AMD Radeon RX 7900 XTX · 24GB VRAM | 34B Q4_K_M | ✅ Runs (tight) | 18–30 tok/s |
Shopping for a machine? What hardware do I need to run Code Llama? →
Check your exact rig: open the checker →
22 models share Code Llama's KV-cache geometry — they size memory the same way, so a rig that fits one tends to fit its same-size kin.
…and 10 more.
Explore the whole catalog by architecture: open the graph →