Unverified60% confidenceFactExact time
DeepSeek developed its own high-level language TileLang, used it to rewrite CUDA operators with smaller code size and higher efficiency, and V3 training has largely stopped relying on the NVIDIA ecosystem.
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
7/25/2026
First Seen
Valid until: 10/23/2026
Sources
梁文锋深度自白：DeepSeek的克制哲学、开源策略与AGI路线图
Related Claims
UnverifiedDeepSeek-V3以极低的训练成本实现了顶尖推理能力72% similarVerifiedDeepSeek achieved mathematical and coding capabilities close to top closed-source models at extremely low training cost, and its R1 model attracted global attention upon release.71% similarVerifiedDeepSeek V3 introduced an FP8 mixed-precision training framework during the pre-training phase, reducing training memory usage and communication overhead.71% similarUnverifiedDeepSeek V4后续版本会支持原生多模态，但多模态只是一个组件，主线仍是把AI训练做好并解决持续学习71% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/613786API
curl https://kongchang.com/api/v1/knowledge/claims/613786MCP
get_claim(id=613786)