Unverified50% confidenceBenchmarkExact time
The researchers selected 159 multi-turn BFCL V4 tasks with complete precise replay paired data for the main analysis.
1
Sources
50%
Confidence
Long-term
Relevance
9/24/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAstraPixels 收录了 171 个已编目天体,每个都附带有据可查的事实信息62% similarUnverified最终生成的数据集包含40032个Needle格式样本,覆盖获取天气、设置定时器、开关灯三个工具,每个工具约有130种表达变体,另加40个不使用工具的负样本60% similarUnverifiedThe HybridDeepResearch benchmark contains 380 tool-dependent tasks.60% similarUnverifiedProcessing 41 documents simultaneously requires the model to have strong long-context comprehension capabilities and cross-file cross-validation capabilities.59% similarUnverifiedThe AFTEC database has collected approximately 1,034 metadata entries, spanning from 1971 to the present.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/946030API
curl https://kongchang.com/api/v1/knowledge/claims/946030MCP
get_claim(id=946030)