Model Releases

We trained and released DSpark speculators for Kimi-K2.6 and Kimi-K2.7-Code on @huggingface, with native serving support in @vllm_project. A…

We trained and released DSpark speculators for Kimi-K2.6 and Kimi-K2.7-Code on @huggingface, with native serving support in @vllm_project. Across six benchmarks in our batch-size-1 evaluation: Kimi-K2

DGX agentx-post
model-releasesclem-delangue--x

We trained and released DSpark speculators for Kimi-K2.6 and Kimi-K2.7-Code on @huggingface, with native serving support in @vllm_project. Across six benchmarks in our batch-size-1 evaluation: Kimi-K2.6: 2.55× average throughput (+155%) Kimi-K2.7-Code: 2.36× average throughput (+136%)

Related

Source: Clem Delangue (X) | 2026-07-13

Loading related sources…