Unverified50% confidenceBenchmarkExact time
In the real-world test, GPT-6 Astra identified all 4 artificially planted errors in the documents, achieving a 100% error detection rate.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/11/2026
First Seen
Valid until: 12/10/2026
Sources
Related Entities
Related Claims
UnverifiedA 2023 study found that GPT-4 produced errors in approximately 15–20% of cases when generating code documentation.72% similarUnverifiedGPT-6 Astra made calculation errors when filling out the U.S. federal individual income tax Form 1040 during an OpenAI demonstration.67% similarUnverifiedGPT-5.4及更早版本在提取文档中的数字时频繁出错，包括小数点位置偏移、千分位分隔符误读、负号遗漏等问题66% similarUnverifiedIn the tax analysis scenario, GPT-6 Astra proactively raised approximately 7 clarifying questions before starting and ultimately delivered a report containing a complete transaction ledger of 3,739 rows.64% similarUnverifiedIn a cancer imaging recognition test, GPT-5.6 misidentified nearly all 6 scans containing tumors, classifying them as hemorrhages or incorrect types.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/897370API
curl https://kongchang.com/api/v1/knowledge/claims/897370MCP
get_claim(id=897370)