Who hosts uncensored models

Three of the seven providers in this directory host a current frontier open-weight model uncensored, and Silk Compute hosts all four in the grid. Each cell gives the input and output price per 1M tokens.

Updated Prices read November 18, 2026

ProviderKimi K3.1 ProMoonshot AI · 2,780BDeepSeek V4.1 FlashDeepSeek · 763BGLM-5.3Z.ai · 753BGLM-5.3 FlashZ.ai · 320B total, 18B active
Silk Compute$3.50 / $8.00$0.24 / $0.90$2.00 / $4.00$0.24 / $1.00
AbliterationNot served$1.00 / $2.00Abliterated Large Flash$3.00 / $5.00Abliterated Large v2Not served
MorphComing soon$0.30 / $1.20DeepSeek V4.1 Flash UncensoredApprovalNot servedNot served
Unfil.AINot servedNot servedNot servedNot served
FeatherlessNot servedNot servedNot servedNot served
Awan LLMNot servedNot servedNot servedNot served
ModelsLabNot servedNot servedNot servedNot served
Providers serving it1321

Served, input / output per 1M tokensListed as coming soonNot served

Who serves each model

What the other four serve

None of these serves a grid model; this is the largest uncensored model each lists.

ProviderLargest uncensored modelMethodContext
Unfil.AIHermes 3 405BFine-tune128,000
FeatherlessQwen2.5 72B AbliteratedAbliterated32,768
Awan LLMLlama 3 8B AbliteratedAbliterated8,192
ModelsLabuncensored-chatUncensoredNot stated

Sources: each provider's pricing and model pages, linked from its profile.