Unverified50% confidenceBenchmarkExact time
In a visual recognition test of 15 receipts, GPT-6 Luna made errors on three, while Haiku 5.5 and DeepSeek V4.1 Flash answered all of them correctly.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/10/2026
First Seen
Valid until: 1/8/2027
Sources
Related Entities
Related Claims
UnverifiedIn the head-to-head comparison among GPT-5.6 Luna, Grok 4, and DeepSeek Flash, Luna wins on quality and safety, while DeepSeek has an advantage in cost and throughput.69% similarUnverifiedIn a Chinese writing test (creating the first chapter of the long novel "Xiuzhen Milk Tea Shop"), Haiku 5.5 performed the best among the three lightweight models, but still fell short of the flagship-level GPT-6 Astra.68% similarUnverifiedModel names such as GPT-6 Luna and their performance data have not yet been independently verified by third parties, and the content comes from a single demo source.67% similarUnverifiedThe per-task cost of GPT-6 Luna Max is reportedly already one-quarter that of DeepSeek V4.1 Flash.67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/993465API
curl https://kongchang.com/api/v1/knowledge/claims/993465MCP
get_claim(id=993465)