Mixtral 8x7B on NVIDIA GeForce RTX 4070 · 12GB VRAM

🐢 CPU-only (slow) — Mixtral 8x7B @ Q3_K_M
Mixtral 8x7B 8x7B (MoE) at Q3_K_M needs ~23.3 GB (weights 21.3 GB + KV 1.0 GB + overhead 1.1 GB @ 8K ctx) of your 10.8 GB usable VRAM — runs on CPU only — expect slow generation.
context @ Q3_K_M: runs to its full 32K
quantneedsspeed
✗ Q8_049.5 GB—
✗ Q5_K_M33.5 GB—
✗ Q4_K_M28.7 GB—
~ Q3_K_M23.3 GB—
$runlocal install mixtral:8x7b
⚠ NVIDIA support is best-effort in v0.1 — verify before relying on it.
⚠ This model spills entirely to system RAM — single-stream speed is governed by your RAM bandwidth (not your GPU's), so it isn't estimated here.
runlocal verdict card
download card share on x share on reddit drop it in your model-drop post
Mixtral 8x7B: CPU-only (slow)
for your own README — links back here

Same model, other machines: Apple M1 · Apple M1 Pro · Apple M2 · Apple M3 Pro · Apple M2 Pro · Apple M4 Pro

Also runs on a NVIDIA GeForce RTX 4070: Qwen3 · DeepSeek-R1 (distill) · Qwen2.5 · Llama 3.1 · Gemma 4

Best models for a NVIDIA GeForce RTX 4070 · 12GB VRAM →

Architecturally similar

These share Mixtral 8x7B's KV-cache geometry — they size memory the same way, so here's how they fit on a NVIDIA GeForce RTX 4070 · 12GB VRAM too:

modelsharesverdict here
Qwen2.5-Coder 8kv/128hd ✅ Runs (tight)
Llama 3.2 8kv/128hd ✅ Runs comfortably
Llama 3.3 8kv/128hd ❌ Won't fit
Mistral 7B 8kv/128hd ✅ Runs (tight)
Mistral Small 3 8kv/128hd ⚠️ Partial GPU offload
Mistral Nemo 8kv/128hd ✅ Runs (tight)
Codestral 8kv/128hd ⚠️ Partial GPU offload
Llama 3.2 Vision 8kv/128hd ✅ Runs comfortably

…and 7 more — see the Mixtral 8x7B page.

Check any combo yourself: open the checker →