Gemma 2 27B
Gemma 2 27B is an open-weights language model from Google, released June 24, 2024. It has 27B parameters (dense). The context window is 8K tokens. On the Epoch Capabilities Index it scores 118.6, rank 189 of 256 rated models.
Check it against your own PC →
Can you run Gemma 2 27B locally?
The smallest common GPU it runs on is 16 GB of VRAM (with 32 GB of RAM): Q3_K_M, split GPU / CPU, roughly 5 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 | Too large | – | – | – | – |
| 8 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 7.2 GB | 10.0 GB | ~3 |
| 12 GB VRAM | Very slow | IQ3_XXS (noticeable loss) | 11.2 GB | 6.0 GB | ~4 |
| 16 GB VRAM | Runs, slowly | Q3_K_M (okay) | 15.2 GB | 4.9 GB | ~5 |
| 24 GB VRAM | Runs on GPU | IQ4_XS (good) | 21.2 GB | – | ~37 |
| 32 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 29.1 GB | – | ~46 |
Ready-made quantizations on Hugging Face: tensorblock/gemma-2-27b-GGUF
Strengths by category
ModelTape's 0–100 fit on Epoch AI benchmark results.
| Category | Score | Based on |
|---|---|---|
| Math | 10.1 | MATH level 5, OTIS Mock AIME 2024-2025 |
| Knowledge | 26.7 | GPQA diamond, MMLU |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| MMLU | 67.6% |
| MATH level 5 | 27.9% |
| GPQA Diamond | 16.0% |
| OTIS Mock AIME | 1.3% |
Links
Hugging Face: unsloth/gemma-2-27b · Epoch AI benchmarks
Models with a similar score
Qwen2-72B · GPT-4 (Mar 2023) · Llama 3-70B · DeepSeek-V2 (MoE-236B, May 2024) · Gemini 1.5 Flash (May 2024) · GPT-4 (Jun 2023) · Mistral Large · Claude 3 Sonnet
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