Is the $6 Million Training Cost of DeepSeek R1 Real?

DeepSeek R1 is reshaping AI with top rankings in benchmark tests and groundbreaking innovations. Its open-source approach is setting new standards, offering unmatched performance and efficiency.

"AI Disruption" publication New Year 30% discount link.

---

DeepSeek R1 has truly broken through the boundaries! After a few days of building momentum, the shockwave from R1 seems to have reached its peak today. The timeline is flooded with discussions. To summarize, there are two interesting points to share: First, R1 has withstood multiple authoritative benchmark tests. Second, R1 has sparked a lot of gossip and reactions, and of course, some perks, because R1's Sam Altman announced that O3 Mini is now available for free (with query limits).

Authoritative Benchmark Tests

Three images undeniably demonstrate the power of DeepSeek R1:

DeepSeek R1 has passed the test in large model arenas like Chatbot Arena, as well as two other independent benchmarks that could not be pre-hacked (Artificial-Analysis, HLE).

Large Model Arena Ranking

The DeepSeek-R1 ranking in the Large Model Arena has been released, and it’s in the top three! Its performance is comparable to the top models. LM Arena (formerly lmsys.org) has highlighted DeepSeek-R1’s impressive performance, calling it an “incredible milestone and gift” to the community.