basetenlabs/deepseek-ultrafast
Rust · ★ 191 · 18d old
- Stars158
- LanguageRust
- LicenseOther
Rust · ★ 191 · 18d old
An ultra-fast inference engine for DeepSeek-V4.1-Flash for sm_100a and sm_103a. It supports 1M context length and multi-server KV cache sharing.
Decode, step time and acceptance length are measured on "implement the tetris game. output HTML code directly - no explanation." Each decode step drafts a block of tokens and verifies it; the acceptance length is the average number of tokens a step emits. Instructions for reproducing the results are at the bottom of this page.
Description and README excerpt from basetenlabs/deepseek-ultrafast on GitHub ↗.
basetenlabs/deepseek-ultrafast was first seen on GitHub · Rising today on Nov 14, 2026. Since then it has appeared in 62 checks on 2 different days. Its best position was #9 of 30, reached on Nov 15, 02:01 UTC. It dropped off the list after Nov 15, 08:01 UTC.
Position on GitHub · Rising today at every check (lower is better).
Best rank per day (UTC).
Platform metric as reported by GitHub.
Archived daily, weekly and monthly lists that include this entry.
Not at the moment. It was last on the list on Nov 15, 08:01 UTC, after 2 days of trending.
#9 on GitHub · Rising today, on Nov 15, 02:01 UTC. Trendoday checks this list every 60 minutes.
It was first seen on Nov 14, 2026 and has been on the list on 2 days (62 checks) since.
Mostly Rust; it is licensed under Other. It has 158 stars and 4 forks.