← Leaderboard
Shoehorn tracking
Quantize a BF16 GGUF with an imatrix to exactly fit your VRAM, then run it with llama.cpp. Every spare megabyte spent where it buys the most quality
notactuallytreyanastasio.github.io ↗ · Show HN thread ↗
Staying power
45/100
Launched
2026-08-1863 days ago · Models & Training
Launch (HN)
99 points14 comments
Score breakdown
Still online20/20
Hacker News mentions after launch6/20
X posts (last 7 days)0/15
Recent activity12/25
Adoption4/15
Launch-day hype3/5
Later mentions on HN (14+ days after launch): n/a (shared host) · X posts linking to it, last 7 days: 0 (as of 2026-10-21) · Up-check HTTP status: 200