Unverified50% confidenceFactExact time
中国的DeepSeek在2025年初以极低成本训练出性能媲美顶尖模型的作品，对西方实验室的大规模训练路线构成范式挑战
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/26/2026
First Seen
Valid until: 11/24/2026
Sources
Related Entities
Related Claims
UnverifiedLeveraging its MoE architecture and efficient training methods, DeepSeek trained a product in 2024 at relatively low cost that rivals top-tier models.79% similarUnverifiedDeepSeek 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.69% similarUnverifiedDeepSeek is known for its open-source approach and low-cost training strategy.69% similarUnverifiedDeepSpeed and Megatron-LM are mainstream large model training frameworks.68% similarUnverifiedThe tutorial recommends topping up as little as 1 yuan (Chinese yuan) to test the DeepSeek API, suggesting costs are very low68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/802189API
curl https://kongchang.com/api/v1/knowledge/claims/802189MCP
get_claim(id=802189)