Unverified50% confidenceFactExact time
Repo65B's $32 million R&D cost covers data, experiments, pretraining, and post-training; the local device mentioned is the DGX Spark and cannot be extrapolated to mean it runs on an ordinary laptop.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/8/2026
First Seen
Valid until: 2/6/2027
Sources
Related Entities
Related Claims
UnverifiedA six-device configuration of four DGX Sparks and two RTX 6000Ds can locally deploy DeepSeek V4.1 Flash, with a total cost of approximately $38,000 (around 270,000 RMB).62% similarUnverifiedWith the proliferation of open-source hardware, low-cost sensors, and 3D printing technology, individual developers can replicate bipedal locomotion experiments on a budget of a few hundred US dollars that previously required tens of thousands of dollars.60% similarUnverifiedDeepSeek V3, released on December 26, 2024, has 671 billion total parameters and 37 billion activated parameters, uses FP8 training, reports a compute cost of $5.576 million in its paper, and matches GPT-4o in performance.57% similarUnverifiedThe reinforcement learning phase of DeepSeek RE cost about 294,000 US dollars, but was built on top of a V3 base of about 6 million US dollars.57% similarUnverifiedTooling for hardware products typically costs tens of thousands to hundreds of thousands of dollars, and it is a one-time investment that is hard to recoup.56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1147353API
curl https://kongchang.com/api/v1/knowledge/claims/1147353MCP
get_claim(id=1147353)