Unverified50% confidenceBenchmarkExact time
In the Shelf Scanner evaluation, Claude Sonnet and Gemini Flash could read almost all identifiable book titles without hallucinating, while smaller models missed more titles and hallucinated more.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/1/2026
First Seen
Valid until: 12/30/2026
Sources
Related Entities
Related Claims
Unverified作者认为从Codex或Claude Code转过来的用户会对Gemini Flash的迭代速度感到震撼60% similarVerifiedGemini 2.5 Flash adopts a Mixture of Experts (MoE) architecture, activating only a subset of parameters during inference.60% similarUnverifiedIn the show, the response quality of the on-device model was noticeably inferior to that of the cloud-based Gemini Flash, with the on-device model tending to dump the entire context all at once and being very verbose.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/964697API
curl https://kongchang.com/api/v1/knowledge/claims/964697MCP
get_claim(id=964697)