Mistral-Small-3.2-24B-Instruct-2506 vs gemma-3-27b-it: VRAM and speed for running locally
Side by side: how much memory each model needs at Q4, Q8 and FP16, how the KV cache grows with context, the cheapest GPU and Mac that run each, and estimated speed on 11 common machines.
Mistral-Small-3.2-24B-Instruct-2506 is lighter: about 17.7 GB at Q4 against 20.1 GB for gemma-3-27b-it, so it runs on 27 tracked devices vs 25. It should also be faster on the same hardware, since it reads 24.0B parameters per token against 27.4B. Which one gives better answers depends on your task, so check benchmarks for that; this page covers what it takes to run them.
| Mistral-Small-3.2-24B-Instruct-2506 | gemma-3-27b-it | |
|---|---|---|
| Maker | mistralai | |
| License | apache-2.0 | gemma |
| Type | Dense | Dense |
| Total parameters | 24.0B | 27.4B |
| Active per token | 24.0B all | 27.4B all |
| Memory at Q4_K_M | 17.7 GB | 20.1 GB |
| Memory at Q8_0 | 30.3 GB | 34.5 GB |
| Memory at FP16 | 56.2 GB | 64.1 GB |
| KV cache growth | 164 MB per 1k tokens, FP16 | not modeled |
| Memory at Q4, 32k context | 22.4 GB 5.4 GB cache | cache not modeled |
| Cheapest GPU at Q4 | Radeon RX 7900 XT $899 | Intel Arc Pro B70 32GB $949 |
| Cheapest Mac / mini PC at Q4 | Mac mini M6 32GB $1,799 | Mac mini M6 32GB $1,799 |
| Cheapest GPU at Q8 | NVIDIA RTX 6000 Ada 48GB $6,799 | NVIDIA RTX 6000 Ada 48GB $6,799 |
| Tracked devices that run it at Q4 | 27 of 82 | 25 of 82 |
Estimated speed on the same hardware
| Device | Mistral-Small-3.2-24B-Instruct-2506 | gemma-3-27b-it |
|---|---|---|
| GeForce RTX 3060 12GB12 GB usable · 360 GB/s · $329 | doesn’t fit | doesn’t fit |
| GeForce RTX 4060 Ti 16GB16 GB usable · 288 GB/s · $499 | doesn’t fit | doesn’t fit |
| GeForce RTX 508016 GB usable · 960 GB/s · $999 | doesn’t fit | doesn’t fit |
| GeForce RTX 309024 GB usable · 936 GB/s · $1,499 | 39–55 | 34–48 |
| Radeon RX 7900 XTX24 GB usable · 960 GB/s · $999 | 40–56 | 35–49 |
| GeForce RTX 509032 GB usable · 1792 GB/s · $1,999 | 74–105 | 65–92 |
| Mac mini M6 24GB18 GB usable · 170 GB/s · $1,299 | tight fit | doesn’t fit |
| Mac mini M6 32GB24 GB usable · 170 GB/s · $1,799 | 5–9 | 5–8 |
| Mac mini M5 Pro 64GB48 GB usable · 307 GB/s · $3,199 | 9–17 | 8–15 |
| Framework Desktop (Ryzen AI Max+ 395) 128GB96 GB usable · 256 GB/s · $3,449 | 8–14 | 7–12 |
| MacBook Pro M5 Max 128GB96 GB usable · 614 GB/s · $6,699 | 19–34 | 17–30 |
Hardware that runs only one of them
Run Mistral-Small-3.2-24B-Instruct-2506 but not gemma-3-27b-it: Radeon RX 7900 XT, NVIDIA RTX 4000 Ada 20GB.
Quick answers
- Which needs less VRAM, Mistral-Small-3.2-24B-Instruct-2506 or gemma-3-27b-it?
- Mistral-Small-3.2-24B-Instruct-2506: about 17.7 GB at Q4_K_M, against 20.1 GB for gemma-3-27b-it. At Q8_0 it's 30.3 GB vs 34.5 GB.
- Which is faster locally, Mistral-Small-3.2-24B-Instruct-2506 or gemma-3-27b-it?
- Decode speed is limited by how many bytes are read per token. Mistral-Small-3.2-24B-Instruct-2506 reads 24.0B parameters per token and gemma-3-27b-it reads 27.4B, so Mistral-Small-3.2-24B-Instruct-2506 should be faster on the same machine. Our estimate on the GeForce RTX 3090 at Q4 is 39–55 tok/s vs 34–48 tok/s. These are estimates, not benchmarks.
- What is the cheapest GPU for Mistral-Small-3.2-24B-Instruct-2506 and gemma-3-27b-it?
- At Q4: Radeon RX 7900 XT ($899) for Mistral-Small-3.2-24B-Instruct-2506; Intel Arc Pro B70 32GB ($949) for gemma-3-27b-it. Prices are dated list-price snapshots.
- Can a Mac run Mistral-Small-3.2-24B-Instruct-2506 and gemma-3-27b-it?
- The cheapest Mac or mini PC we track for Mistral-Small-3.2-24B-Instruct-2506 is the Mac mini M6 32GB ($1,799). For gemma-3-27b-it it's the Mac mini M6 32GB ($1,799).
- Which is better, Mistral-Small-3.2-24B-Instruct-2506 or gemma-3-27b-it?
- We don't score answer quality. This comparison covers memory, context cache, hardware and speed. Pick the one that fits your hardware, then test both on your own prompts.
Related comparisons
- Gemma 4 31B vs gemma-3-27b-it22.8 vs 20.1 GB
All comparisons · MoE vs dense · Q4 vs Q8 vs FP16 · Context length and VRAM · VRAM calculator