Kimi K3 Hits 423 tok/s with SGLang Integration

Kimi K3 has achieved Day-0 support on SGLang, reaching 423 tok/s on the GSM8K benchmark by utilizing a DSpark speculator draft model trained with SpecForge to optimize decoding throughput.

2026-07-28 ~ 2026-07-28 · 4 related posts

• Kimi K3 launches on SGLang with 423 tok/s and 11 cloud partners — ying11231 · 2026-07-28
• SGLang adds day-one Kimi K3 support and reports 423 tokens/s on GSM8K — BanghuaZ · 2026-07-28
• Kimi K3 serving stack reaches 423 tok/s after DSpark draft-model tuning — ying11231 · 2026-07-28
• SGLang Day-0 Support for Kimi K3 Hits 423 tok/s on GSM8K — ying11231 · 2026-07-28