DeepSeek V4 Pro on Together AI Ranks No. 1 on Artificial Analysis for Output Speed, Latency

aiai-infrastructureai-inference-platforms 4 posts · 3 accounts

DeepSeek V4 Pro on Together AI is now ranked No. 1 on Artificial Analysis for both output speed and latency, according to Together AI.

Together AI said serving V4 well is an inference-systems problem involving KV cache, prefix reuse, kernels and endpoint profiles, and tied the ranking to that systems work.

From the sources (4 posts)

@vipulved

DeepSeek V4 Pro on @togethercompute becomes #1 on both latency and speed.

@tri_dao

RT @vipulved: DeepSeek V4 Pro on @togethercompute becomes #1 on both latency and speed.

@togethercompute

RT @vipulved: DeepSeek V4 Pro on @togethercompute becomes #1 on both latency and speed.

@togethercompute

DeepSeek V4 Pro on Together AI is now #1 on Artificial Analysis for both output speed and latency. Serving V4 well is an inference systems problem: KV cache, prefix reuse, kernels, and endpoint profiles. We break down the systems work her

Kimi Code Adds One-Line Install, Video Context in Open-Source Coding Agent Upgrade NVIDIA Research Releases MotionBricks Open Model for Character Animation at 15,000 FPS