Unverified50% confidenceBenchmarkExact time
Qwen2.5-72B approaches GPT-4 level performance on benchmarks such as MMLU and HumanEval.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/6/2026
First Seen
Valid until: 12/5/2026
Sources
Related Entities
Related Claims
UnverifiedQwen2.5-27B是阿里云通义实验室发布的270亿参数开源模型，在MMLU、HumanEval、GSM8K等基准上接近GPT-4早期版本水平76% similarUnverified在文档摘要基准测试上，Qwen2.5-32B-Instruct和Llama-3.1-70B的量化版本已能达到GPT-4早期版本85%-92%的表现水平76% similarUnverifiedOn mainstream benchmarks such as MMLU, HumanEval, and GSM8K, open-source models have reached 85–95% of the level of early versions of GPT-4.75% similarUnverifiedQwen2.5-Coder demonstrates performance approaching GPT-4o on code generation benchmarks such as HumanEval, and its 7B version can run on an ordinary laptop after quantization.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/864374API
curl https://kongchang.com/api/v1/knowledge/claims/864374MCP
get_claim(id=864374)