qwen2.5 instruct Benchmark | LLM Performance Benchmarks