Unverified50% confidenceBenchmarkExact time
GPT-5.4 was reversed by a "trusted source" on 44.7% of questions, Grok-4.20 had a reversal rate as high as 87.5%, while Gemini-3.1-Pro was nearly unaffected at just 0.6%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/1/2026
First Seen
Valid until: 12/30/2026
Sources
Related Entities
Related Claims
UnverifiedEarly GPT-4 achieved a resolution rate of less than 2% on SWE-bench, while recent top models have surpassed 50%.78% similarUnverifiedGrok 4.5 achieved 83.3% on the Terminal Bench benchmark, roughly on par with GPT 5.5 and slightly below Claude 5.77% similarUnverifiedTop autoregressive models such as GPT-4o and Claude 3.5 have already exceeded 95% accuracy on GSM8K, leading to benchmark saturation.77% similarUnverifiedGPT-5.6's coding capability reached 91.9% on key benchmarks and its safety evaluation CyberGym score reached 85.6%, both surpassing competitor Nexus 5.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/965811API
curl https://kongchang.com/api/v1/knowledge/claims/965811MCP
get_claim(id=965811)