Runs on ≤16 GB · Hugging Face ↗
Not independently scored by Artificial Analysis — its base model Qwen/Qwen2.5-14B is the closest proxy
NetEase Youdao's 14B on Qwen2 (apache-2.0), with its own GGUF.
BigCodeBench-Hard pass@1 19% (28/148), via official protocol · llama.cpp Q6_K · one request at a time · a shared cluster GPU · mean of 3 runs, range 0.7 pp. The brutal counterpart to HumanEval — where HumanEval saturates near the top, BCB-Hard spreads the field, so this is the number that actually separates coding ability.
Our own runs on this model. Each number carries how many problems it is over and how many times we ran it — these are not comparable with each other, and none of them is on the same scale as the third-party indices on the leaderboard.
| test | score | over | conditions |
|---|---|---|---|
| BigCodeBench-Hard | 18.7% | n=148 · 3 run(s) | official protocol · llama.cpp Q6_K · one request at a time · a shared cluster GPU · mean of 3 runs, range 0.7 |
Real GGUF file sizes = weight VRAM. Add the KV cache for your context (≈6 GB at 32K, fp16). Read from the GGUF header and checked against what llama.cpp actually allocates.
| Quantization | Size |
|---|---|
| Q5_K_M | 10.5 GB |
| Q6_K | 12.1 GB |
| F16 | 29.5 GB |
18.9% over three runs (18.9 / 18.2 / 18.9), and it is the only one in this batch we measured at Q6_K rather than Q4_K_M, so compare sizes before compare scores.