Unverified50% confidenceFactExact time
The first complete baseline run reported all 176 tests passing while appending 30 lines of synthetic data to the production log.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/12/2026
First Seen
Valid until: 1/10/2027
Sources
Related Entities
Related Claims
UnverifiedThe test ultimately produced a sample of 174 tasks and 1,744 attempts, using an A/B comparison method with the same routing, same platform, and same time period.74% similarUnverifiedThe fine-tuning used 4,260 training samples and 474 evaluation samples, all of which were STEM question-and-answer data with chain-of-thought.67% similarUnverifiedThe author reviewed approximately 1,500 detection calls, manually mined false positives, adjusted bounding boxes, and then retrained the model.66% similarUnverifiedThe test ran a total of 130 variants, covering different levels of reasoning effort, called via OpenRouter with official endpoints for each lab locked in wherever possible.65% similarUnverifiedCodex使用了126个子智能体、32500次工具调用、391次浏览器测试和2300多次单元测试；Claude Code使用35个子智能体、102次浏览器测试和296次单元测试65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1000656API
curl https://kongchang.com/api/v1/knowledge/claims/1000656MCP
get_claim(id=1000656)