Unverified50% confidenceBenchmarkExact time
In a test of 360 clean texts, Jev made only 1 misjudgment, gpt-5-mini made 1 misjudgment, Gemini 2.5 Flash-Lite and Gemma each made 2 misjudgments, gpt-oss-safeguard-20b made 4 misjudgments, and Llama Guard 4 made 26 misjudgments.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/13/2026
First Seen
Valid until: 10/27/2026
Sources
Related Entities
Related Claims
UnverifiedWith two to three attempts, GLM-5.3 Flash scored 82%, which is within the margin of error of GPT-5.6 Luna's 81%.68% similarVerifiedOn mainstream benchmarks such as MMLU and HumanEval, the score gaps among Claude 3.5, GPT-4o, Gemini 1.5 Pro, and Grok-2 have been compressed to within the range of statistical error.67% similarUnverifiedIn the PR split-review example, GPT-6.1 Sol consumed input=77, output=266, and reasoning=149 tokens, took 5.4 seconds, did not fabricate diffs, and did not claim to have run tests.66% similarUnverifiedQwen3-235B performed the worst on injection tests, failing 5 out of 8 attacks; Gemini 3.7 Flash, Claude Haiku 5.5, and Claude Opus 4.5 achieved zero failures across all 8 injection attacks.66% similarUnverifiedGemini 3.7 Flash answered all 169 returned items correctly in the sufficiency task, but 23 calls yielded no result after three attempts; counting those as failures reduces the number of passing families from a perfect score to 28 out of 48.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1015386API
curl https://kongchang.com/api/v1/knowledge/claims/1015386MCP
get_claim(id=1015386)