Unverified75% confidenceFactTime unknown

Qwen3.5在多模态视觉推理上首次反超了Gemini 3 Pro和GPT-5.2

1

Sources

75%

Confidence

Medium-term (~90 days)

Relevance

6/1/2026

First Seen

Valid until: 8/30/2026

Sources

Qwen3.5深度解析：混合注意力架构实现19倍长上下文加速

bilibili雪天鱼

Related Entities

Qwen3.5model Gemini 3 Promodel GPT-5.2model

Related Claims

VerifiedGPT-5.5在多项能力上全面超越Claude 4.7和Gemini 3.1 Pro等模型78% similar Verified目前主流的Agent底座模型包括GPT-4o、Claude 3.5 Sonnet、Gemini 1.5 Pro等闭源模型，以及Llama 3、Qwen2.5、DeepSeek等开源模型72% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/18079

API

curl https://kongchang.com/api/v1/knowledge/claims/18079

MCP

get_claim(id=18079)