Expired50% confidenceOpinionExact time

There is speculation that DeepSeek V4 may have distilled closed-source proprietary models, as its outputs are sometimes strikingly similar to those of closed-source flagship models.

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/23/2026

First Seen

Valid until: 10/21/2026(expired)

Sources

Three Major Chinese AI Models Incoming: Qwen 4, DeepSeek V4, and GLM Next Version Breakdown

Related Claims

UnverifiedThis comparative experiment used TRAE + DeepSeek V4 Pro versus Claude Code + DeepSeek V4 Pro, with both sides using the same model, DeepSeek V4 Pro (1M context window).74% similar UnverifiedDeepSeek's parent company has already adopted V4 as its daily programming model internally.73% similar UnverifiedTechnically, Claude Code can be connected to other models such as DeepSeek V4 Pro or GLM 5.1, but the gap in effectiveness compared to the Opus model is enormous.73% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/597086

API

curl https://kongchang.com/api/v1/knowledge/claims/597086

MCP

get_claim(id=597086)