Unverified50% confidenceBenchmarkExact time
Scores on the MATH benchmark surged from approximately 50% to over 90% between 2023 and 2024, and top models on GSM8K have already approached a perfect score.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/8/2026
First Seen
Valid until: 11/6/2026
Sources
Related Claims
UnverifiedSelf-Consistency can bring improvements of 5–20 percentage points on mathematical reasoning benchmarks such as GSM8K.73% similarVerifiedIn 2024, top models achieved a pass rate of approximately 50–70% on SWEbench-Verified.72% similarUnverifiedIn early 2023, mainstream models' SWE-bench scores were generally below 5%; by mid-2025, top models had surpassed 70%.70% similarUnverifiedThe solve rate of top models on SWE-bench jumped from under 5% in 2023 to over 30% in 2025.66% similarUnverifiedThe per-step accuracy of current top-tier models is approximately between 90% and 95%.63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/706719API
curl https://kongchang.com/api/v1/knowledge/claims/706719MCP
get_claim(id=706719)