Unverified50% confidenceBenchmarkExact time
In the easy-to-verify category, seven models scored a perfect 40/40 across all six bait types, and Claude Sonnet 5.5, Gemini 3.7 Flash, and GPT-6.1 Sol answered every question correctly across all six tasks in both the easy and hard categories.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/11/2026
First Seen
Valid until: 1/9/2027
Sources
Related Entities
Related Claims
UnverifiedThe model detection tool can complete detection of approximately 28 models in three minutes, with a supported model pool that includes Opus 5, Opus 5.5, GPT-6 Astra, Sonnet 5, and others.70% similarUnverifiedIn a test of identifying hidden bugs in images, Sonnet 5's Max mode spent approximately 10 minutes and identified all bugs with completely correct results, while Gemini's top-tier model only found 4, and GPT-5.5 X-High failed to detect any.69% similarUnverifiedAmong 8 locally run large models tested on a Raspberry Pi 5, only 3 genuinely passed the tests for speed and task completion.66% similarUnverifiedIn the hard-to-verify category with injection bait, the number of questions for which GPT-5.4 mini judged incorrectly in both orderings rose from 3 to 39, Qwen3-Next 80B Instruct from 4 to 38, and GPT-5.4 mini and Qwen3-Next Instruct scored zero on hard questions (below the approximately 0.25 expected from random coin flipping).64% similarUnverifiedBased on comprehensive empirical test results, the capability ranking of the four models is: Sol > GPT-5.5 > Terra > Luna.64% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/996774API
curl https://kongchang.com/api/v1/knowledge/claims/996774MCP
get_claim(id=996774)