Unverified50% confidenceFactExact time
Frontier Math 是由 EpochAI 创建的数学基准测试，此前 AI 模型在该测试上的通过率不足 2%
1
Sources
50%
Confidence
Long-term
Relevance
8/24/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedOpenAI's chart shows that at lower test-time compute levels, the model's success rate on this problem is close to zero, with occasional success only under extremely long reasoning chains.63% similarUnverifiedDevin 2.0 resolved 13.86% of real-world programming problems in the SWE-bench benchmark, compared to only 1.96% for previous AI models.63% similarUnverifiedThe solve rate of current top AI models on SWE-bench remains far below the level implied by their competition performance.60% similarUnverifiedMultiple independent studies in 2023 showed that some top models scored extremely high on standard mathematical reasoning tests but performed significantly worse on equivalent rephrased versions, indicating that the models learned test patterns rather than genuine reasoning ability.60% similarUnverifiedIn the Agency tool-calling test, both models scored 95 points, and the only failure for both was task No. 13 (USD to JPY foreign currency conversion).58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/792804API
curl https://kongchang.com/api/v1/knowledge/claims/792804MCP
get_claim(id=792804)