Expired50% confidenceBenchmarkExact time
Using V4 Pro with Thinking reasoning mode in the Claude Code environment for a T5 test took approximately 6 minutes and 48 seconds, and the generated page had a layout bug where footer elements appeared in the upper-left corner.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
8/26/2026
First Seen
This claim has expired. It is kept for the record and no longer describes the current state.
Valid until: 9/9/2026(expired)
Sources
Related Entities
Related Claims
UnverifiedIn the BugFind code bug-finding test, Qwen3.6 27B Q4 precision scored 87 points, higher than Q6's 80 points, exhibiting a quantization reversal phenomenon.63% similarUnverifiedIn actual testing using the Qwen3 27B Q4 quantized dense model, the generation speed with thinking mode disabled is approximately 58–60 tokens/s.61% similarUnverifiedIn the real-world test, both sides used Claude terminal with DeepSeek v4 and executed the same prompts.60% similarUnverifiedTesting confirmed that the 2048 game generated by Claude Code runs correctly with no obvious bugs: the 4×4 board, number sliding, collision merging, and score calculation all function properly.59% similarUnverifiedV4 Pro Max满血推理模式下算法竞赛3206分、编程93.5分、科学问答90.1分、软件工程80.6分、终端操作67.9分58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/801978API
curl https://kongchang.com/api/v1/knowledge/claims/801978MCP
get_claim(id=801978)