Unverified50% confidenceBenchmarkExact time
Devin's initially announced unassisted resolution rate on SWE-bench was approximately 13.86%, far exceeding other solutions at the time.
1
Sources
50%
Confidence
Long-term
Relevance
9/3/2026
First Seen
Sources
Related Entities
Related Claims
Unverified针对金融领域微调的嵌入模型（如FinBERT）相比通用嵌入模型在财报问答任务上准确率高出15%以上55% similarUnverified报告测出DiffusionGemma平均每步生成token数达到19.74个，而自回归方法每步只能生成1个55% similarUnverifiedDistilBERT将参数量压缩至6600万（原版1.1亿），推理速度提升60%，精度损失不足3%55% similarUnverifiedOn SimpleBench, Sol Pro scored 71.7% and Sol approximately 65%, still 10 percentage points below Fable.55% similarUnverifiedWhen SWE-bench was first released, the pass rate of top models was below 2%; the pass rate of leading systems has now surpassed 50%.55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/846967API
curl https://kongchang.com/api/v1/knowledge/claims/846967MCP
get_claim(id=846967)