Unverified50% confidenceBenchmarkExact time
In the new version of the 24-point problem (numbers 2, 7, 8, 10, requiring three different solutions), DeepSeek V4.1 Flash got all three solutions wrong, each violating the constraint that each number can only be used once.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/11/2026
First Seen
Valid until: 12/10/2026
Sources
Related Entities
Related Claims
Unverified盘古2.0 Flash在SWE-Bench Verified工程实战测试中得分仅63.1，在所有对比模型中垫底64% similarUnverified盘古2.0 Flash在可验证指令遵循测试中得分95.9，位列全场第一63% similarUnverified盘古2.0 Flash在AIME类数学测试中得分93.3，在HMMT测试中得分91.5，数学能力处于第一梯队62% similarUnverifiedIn three benchmarks—Agent's Last Exam, ZeroBench, and DeepSWE—V4-Flash-Vision-Exp outperformed Opus 4.8.62% similarUnverifiedAmong all local models benchmarked by UP host bwd6, Qwen3.8 Flash Next ranked second, losing by one point to its sibling 27B model using the BF16 unquantized version.60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/908720API
curl https://kongchang.com/api/v1/knowledge/claims/908720MCP
get_claim(id=908720)