Expired50% confidenceBenchmarkExact time
Large language models such as GPT-4 and Gemini Ultra have achieved near top-student performance on competition mathematics, but still make frequent errors on research-level proofs
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/19/2026
First Seen
Valid until: 10/17/2026(expired)
Sources
Related Claims
VerifiedBetween 2023 and 2024, the performance gap among mainstream large language models such as GPT-4, Claude 3, and Gemini Ultra continued to narrow on benchmarks, making them difficult to distinguish on most general tasks.82% similarVerifiedNew-generation large language models such as GPT-4, Claude 3.5 Sonnet, and Gemini 1.5 Pro have achieved significantly higher scores on code capability benchmarks such as HumanEval and SWE-bench compared to previous versions.80% similarVerifiedRepresentative large language models such as GPT-4, Claude, and Gemini typically have parameter counts in the hundreds of billions or more.76% similarUnverifiedMultimodal large models (such as GPT-4V and Gemini) use end-to-end processing to avoid cascading errors from character segmentation mistakes, and often outperform traditional OCR on complex text recognition75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/565060API
curl https://kongchang.com/api/v1/knowledge/claims/565060MCP
get_claim(id=565060)