[Overseas Developers Review Kimi K3: An F1 in the Display Case, Also a Gift to the World] According to monitoring by Beating, after the full weights of Kimi K3 were released, the overseas community gave it high praise, though opinions were not entirely unanimous. Admirers call it a 'monster,' believing it has reached the level of top-tier closed-source models. Interface generation, code architecture, and deep research are currently the most highly praised capabilities. However, some feel it has not yet surpassed Fable 5, though it is catching up quickly. The biggest regret remains its size. A Reddit user compared it to 'an F1 car in a display case': the performance is indeed powerful, but ordinary people simply can't drive it home. Some developers have already connected 80 RTX 5090 GPUs into a cluster to run the full Kimi K3. It does not use expensive data center GPUs like H100 or H200, but the entire setup is still beyond the affordability of regular individual users. K3 has also been criticized for being too token-intensive. Some users claim it takes a long time to process and consumes tokens quickly when handling complex tasks. One user reportedly used up their weekly quota in just a day and a half, prompting calls for rankings to consider not only scores but also the number of tokens required to complete the same task. The community is still debating whether it qualifies as truly open-source. Critics argue that K3 has not disclosed its training data and its license comes with restrictions, making 'open weights' a more accurate description. Supporters, however, are less concerned about the terminology: being able to download, modify, and deploy it independently already offers significantly more freedom compared to relying solely on closed-source APIs. [Original Link]