DeepSeek's New Model Is the Cheapest Well-Known AI to Run, Analysis Finds

In one line: A research firm reportedly found that DeepSeek's new AI model is by far the cheapest to run among well-known models.

Key points

• Per Reuters, a research firm's analysis concluded that the inference (run) cost of DeepSeek's new model is far lower than that of well-known rivals.
• Lower run costs mean the same workload can be served with less compute and expense, a gap that widens at deployment scale.
• DeepSeek, a China-based AI developer, has previously drawn attention for its low-cost training and operating approach.

Why it matters

Run cost is a primary driver of the real bill for operating a model in production. If a best-in-class inference price holds up, unit economics — not just benchmark performance — could weigh more heavily in enterprises' model-adoption decisions.

More

• DeepSeek's new AI model is by far the cheapest of well-known models to run, research firm says — Reuters