StarCoder 2 15B
StarCoder 2 15B is an open-weights language model from Hugging Face, released February 20, 2024. It has 15B parameters (dense). On the Epoch Capabilities Index it scores 96.3, rank 224 of 256 rated models.
Check it against your own PC →
Lab
Hugging Face
Released
February 20, 2024
Weights
Open restricted use
Parameters
15B dense
Context
–
Price / M tokens
–
Epoch ECI
96.3 #224
Can you run StarCoder 2 15B locally?
The smallest common GPU it runs on is 6 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 | Runs, slowly | Q3_K_M (okay) | 5.2 GB | 5.4 GB | ~5 |
| 8 GB VRAM | Runs, slowly | IQ4_XS (good) | 7.2 GB | 4.0 GB | ~6 |
| 12 GB VRAM | Runs on GPU | IQ4_XS (good) | 11.2 GB | – | ~26 |
| 16 GB VRAM | Runs on GPU | Q5_K_M (excellent) | 13.9 GB | – | ~27 |
| 24 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 15.5 GB | – | ~44 |
| 32 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 15.5 GB | – | ~80 |
Benchmark results
| Benchmark (Epoch AI) | Score |
|---|---|
| MMLU | 52.1% |
Links
Models with a similar score
DeepSeek-Coder-V2-Lite-Base · Falcon 2 11B · Nemotron-4 15B · internlm-20b · PaLM 2-M · Qwen2.5-Coder-3B · Llama 2-13B · LLaMA-33B