It is worth anticipating the true performance of official DeepSeek V4. While the lack of multimodal capabilities results in subpar front-end presentation, it has demonstrated near state-of-the-art competitiveness against leading mainstream models in my series of task benchmark tests.
Judging by current performance, both Kimi K3 and DeepSeek V4 deliver highly competitive capabilities, and GLM-5.3 is also well worth looking forward to.
Judging by current performance, both Kimi K3 and DeepSeek V4 deliver highly competitive capabilities, and GLM-5.3 is also well worth looking forward to.