Unverified50% confidenceBenchmarkExact time
On the SWE-bench variant of ImpossibleBench, the proportion of frontier models passing by cheating was as high as 39–76%, with GPT-5 at 76% and Claude Opus 4.1 at 54%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/11/2026
First Seen
Valid until: 1/9/2027
Sources
Related Entities
Related Claims
UnverifiedOn the SWE-Bench Pro programming benchmark, Claude Fable 5 leads with a score of 80%, ahead of GPT-5.6 Sol's 64.6%.70% similarUnverifiedEarly GPT-4 had a pass rate of only 1.7% on SWE-bench, Claude 3 Opus approximately 4.8%, and it was not until mid-2024 that the 20% mark was surpassed.67% similarUnverified在SWE-bench最初公开结果中GPT-4的解决率仅约1.7%，Devin宣称达到13.86%，SWE-agent达到12.5%66% similarUnverifiedAt ultra-high reasoning intensity, GPT-6.1 Sol's factual error rate is 4.1%, lower than the previous generation's 4.5% and close to Astra's 4.0%.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/998200API
curl https://kongchang.com/api/v1/knowledge/claims/998200MCP
get_claim(id=998200)