DeepSeek and MiniMax Going Global Is the Story the Leaderboards Missed
DeepSeek and MiniMax got Chinese approval to deploy globally this week, and I think it matters more than another benchmark record. Approval like that is a door, and these labs just walked through it.
The West watches Chinese models on leaderboards and shrugs. But a model you can actually deploy, in a language you actually speak, at a price that undercuts the incumbents, is a product, not a score.
DeepSeek's V4.1 Flash already beats Meta's Muse Spark Open on speed and cost in local tests. That is the combination that stings: open, fast, and cheap. The closed frontier is no longer the only serious option.
I am not picking sides. I am noting that the 'best model' question now has answers that do not all come from the same three companies. Competition like that is good for everyone who pays for tokens.
