⌦abliteratedmodels.org

Abliterated models for 24 GB VRAM

Abliterated models that fit in 24 GB of GPU memory at the Q4_K_M quantisation with a 4,096-token context — RTX 3090, 4090 and comparable cards. 712 models qualify, the largest being 36B parameters at roughly 22.7 GB. Estimates are derived from parameter counts, not measured.

712 models↓ 49.2M downloads

Need a figure for your exact machine, including system RAM, a different quantisation or a longer context? Use the hardware checker.

Qwythos-9B-v2

9BOtheruncensored

from empero-ai/Qwythos-9B-v2

↓ 315K♥ 32411 repos

Qwythos-27B-v1

27BOtheruncensored

from empero-ai/Qwythos-27B-v1

↓ 192.3K♥ 1737 repos

Homura-30B

30BOtherheretic

from darkc0de/Muse-Glimmer-30B-heretic

↓ 85.4K♥ 323 repos

Showing the 120 most-downloaded of 712 models in this group. The remaining 592 are reachable through the full paginated index and are all listed in the sitemap.