Hacker NewspichMon, Aug 17, 2026, 7:29 AM PDT
score 23.6
38HN36HN cmts
Running a 27B AI model at 256K context on a 24GB GPU at 50 tokens per second
Original: Qwen3.8 27B at 256K: 50 TPS on a 24 GB GPU
Source: piszczek.pl ↗
Writing ELI5 summary…
Original: Qwen3.8 27B at 256K: 50 TPS on a 24 GB GPU
Source: piszczek.pl ↗
Writing ELI5 summary…