Gemma 3 27B
Gemma 3 27B is an open-weights language model from Google, released March 12, 2025. It has 27B parameters (dense). On the Epoch Capabilities Index it scores 127.5, rank 164 of 256 rated models.
Check it against your own PC →
Can you run Gemma 3 27B locally?
The smallest common GPU it runs on is 12 GB of VRAM (with 32 GB of RAM): IQ3_XXS, split GPU / CPU, roughly 7 tokens per second. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 5.2 GB | 8.9 GB | ~3 |
| 8 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 7.2 GB | 6.9 GB | ~4 |
| 12 GB VRAM | Runs, slowly | IQ3_XXS (noticeable loss) | 11.2 GB | 2.9 GB | ~7 |
| 16 GB VRAM | Runs on GPU | IQ3_XXS (noticeable loss) | 14.1 GB | – | ~28 |
| 24 GB VRAM | Runs on GPU | Q5_K_M (excellent) | 23.0 GB | – | ~28 |
| 32 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 25.9 GB | – | ~46 |
Strengths by category
ModelTape's 0–100 fit on Epoch AI benchmark results.
| Category | Score | Based on |
|---|---|---|
| Math | 20.4 | MATH level 5, OTIS Mock AIME 2024-2025 |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| MATH level 5 | 74.0% |
| GPQA Diamond | 31.6% |
| OTIS Mock AIME | 22.4% |
| Aider Polyglot | 4.9% |
Links
Models with a similar score
Qwen2.5-Max · Gemini 1.5 Pro (Sept 2024) · Claude 3.5 Sonnet · Grok-2 (Dec 2024) · Phi-4 · GPT-4o (Nov 2024) · GPT-4o (May 2024) · Llama 3.1-405B
More from Google
Gemini 4 Carbon · Gemini 4 Argon · Gemini 3.8 Flash · Gemini 3.7 Flash · Gemini 3.6 Flash · Gemini 3.5 Flash-Lite · Gemini 3.5 Flash Cyber · Gemma 4 E4B