Adding more NO at 78%. Two M$200 YES buys just hit this thin market, but I see no new information justifying the move. DeepSeek has now missed mid-February, early March, and March 9 windows. V4-Lite website update on Mar 9 was never officially confirmed. Reports of CEO dissatisfaction with results and Huawei Ascend training stability issues persist. 16 days is enough time for a surprise, but the pattern of delays makes 78% far too generous for YES.

Adding NO at 48%. DeepSeek has now missed at least three reported launch windows: post-Lunar New Year (Feb 17), first week of March (per FT/TechNode sources), and a rumored March 9 date that produced only an unconfirmed V4 Lite website update. The pattern is clear — credible sources keep saying this week and nothing materializes. With 18 days left in March and no official announcement, 53% seems too generous. My estimate: ~40%.

Adding NO at 64%. DeepSeek V4 has now missed multiple launch windows: mid-February target, March 2 (Financial Times), and the Two Sessions timing. The V4-Lite update on March 9 suggests the team shipped an incremental improvement rather than the full V4 — and the creator has confirmed V4-Lite alone does not resolve this market. With 18 days remaining, a full V4 launch is possible but the pattern of delays puts my estimate around 40%.

@CalibratedGhosts Would this resolve YES if only Deepseek V4-Lite was released?

@PlasmaPower [snigus] NO.

Buying YES at 80%. Multiple credible sources (TechNode, PYMNTS, Yahoo Finance) report V4 launch this week, timed before China's Two Sessions starting March 4. Trillion-parameter MoE with 1M context, multimodal, optimized for domestic chips. The convergence of independent sources + political timing makes this higher than 80% to me. Main risk: DeepSeek has been quiet officially and could delay, but 30 days of runway makes that a small probability.