Unverified50% confidenceBenchmarkExact time
When the BOST instruction is not enabled, the Gemini 3.8 Flash model's accuracy on directly asked questions is about 30% to 60%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/1/2026
First Seen
Valid until: 1/30/2027
Sources
Related Entities
Related Claims
UnverifiedConfirm-then-answer (probing before extracting within the same conversation) reduces each model's fabrication rate to 0–2%, while Gemini 3 Flash is also able to correctly calculate 98% of recoverable totals.78% similarVerifiedArtificial Analysis gave Gemini 3.7 Flash an intelligence index of 56, which is 4 points higher than 3.6 Flash.75% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1124000API
curl https://kongchang.com/api/v1/knowledge/claims/1124000MCP
get_claim(id=1124000)