🎉 Winners Announced: Mars Global Creator Program After thorough review of all outstanding submissions, we’re excited to officially reveal the winners of the Mars Global Creator Program! 🏆 Top 1 – 5 @Meow78600_ @ahmx_web3 @shakibmunsi0 @Race_varen @Joshua_creates 🎖️ Top 6 – 20 @cuemnee @UmarfarukU11588 @Only1maryfaith @GookieNft @GhostInTheDAO @hanhanbtc @zqd1125979 @Defishakil @Ada_grams @Brittney0009 @atiqur2904 @IroIbeGodswill7 @Web3Salvage_ @Confipaul09 @DanniXchange Our admin will DM all remaining winners directly with your individual results shortly. Thank you to every creator who entered — we were truly impressed by the quality and creativity of your work. Huge congratulations to all our winners! 🚀 #MarsGlobalCreatorProgram# #MarsChain# #Web3Creators#

Six weeks ago, serving DSpark in vLLM meant choosing a draft length for your traffic and living with it. Now you set that length once, and vLLM decides how much of the draft to verify every step. ✨ On DeepSeek-V4-Pro-0813, the first token of a 7-token draft survives verification more than 70% of the time. The last one, less than 10%. 📈 One config, adaptive verification on with num_speculative_tokens 7, holds the Pareto frontier from concurrency 1 to 256 on 8×B300. Long draft at low load, short at high load. On main behind enable_adaptive_verification, for DSpark with a confidence head on Flash Attention or DSV4 attention. More backends and models are in bring-up. Thanks to Lucas Wilkinson (@RedHat_AI) and Ben Chislett (@NVIDIAAI), and to @deepseek_ai for DSpark and the varlen DeepGEMM indexer kernel. 🔗