Capital · operating
no revenue disclosed — 'focuses on research, no immediate commercialization plans'; V3 training cost ~$5.576M (official technical report); V4 (Apr-2026) training cost NOT disclosed in any source found this pass — cost-efficiency claims ('50x lower cost', 'fraction of the cost') are relative/marketing, not a dollar figure
Sourced figures 4
- Wikipedia — DeepSeek states it focuses on research, no immediate commercialization plans
- Wikipedia — DeepSeek V3 training cost $5.576M (official technical report; vs claimed $100M for GPT-4)
- Google News RSS — DeepSeek V4 training cost sweep No V4 training-cost figure disclosed; coverage only gives relative claims ('50x lower cost, Claude-level benchmarks')
- Google News RSS — DeepSeek V4 training cost sweep (attrib. Tom's Hardware / SemiAnalysis / The Register) Disputed historical claim: 'DeepSeek might not be as disruptive as claimed, firm reportedly has 50,000 Nvidia GPUs and spent $1.6 billion' (SemiAnalysis-sourced, contested by The Register)
the working
Threads 1
A claim is one metric we assert about a node — the value is the summary, this page is the receipt. Sourced from kestrel; supersedes-history not yet shown. Back to DeepSeek.