Updated November 18, 2026

Uncensored LLM leaderboard.

34 uncensored and abliterated models you can call by API, ranked on how smart the original is, whether they finish agent tasks, whether they answer, how fast they run and how far they drifted.

Short answer

#1 DeepSeek V4.1 Flash 1017 Uncensored (69.8), #2 GLM 5.3 Flash Uncensored (69.7), #3 Qwen 3.8 27B Uncensored Fast by Heabsy (66.8). Scores are 0 to 100: intelligence 40%, agentic 30%, non-refusal 15%, speed 10%, low drift 5%. Every model is live on one OpenAI compatible API.

The leaderboard

All 34 ranked uncensored models.

#ModelScoreIntelligenceAgenticAnswersSpeed$ / 1M in · out
1DeepSeek V4.1 Flash 1017 Uncensoredfrom DeepSeek V4.1 Flash 101769.867.555 (2/5)100%239 tok/s$0.33 · $1.32
2GLM 5.3 Flash Uncensoredfrom GLM 5.3 Flash69.765.566 (3/5)94%178 tok/s$0.12 · $0.50
3Qwen 3.8 27B Uncensored Fast by Heabsyfrom Qwen 3.8 27B66.862.249 (2/5)100%245 tok/s$0.83 · $2.62
4DeepSeek V4.1 Flash 0910 Uncensoredfrom DeepSeek V4.1 Flash66.165.153 (2/5)97%209 tok/s$0.33 · $1.32
5DeepSeek V4.1 Flash Uncensoredfrom DeepSeek V4.1 Flash64.365.140 (1/5)100%77 tok/s$0.12 · $0.45
6Abliterated Large Flashfrom DeepSeek V4.1 Flash 101764.067.538 (1/5)92%184 tok/s$1.00 · $2.00
7MiMo V2.6 Flash Uncensoredfrom MiMo V2.6 Flash63.562.355 (2/5)83%118 tok/s$0.46 · $1.50
8Apex Flash 1 Abliteratedfrom GLM 5.3 Flash63.265.570 (3/5)86%10 tok/s$0.15 · $0.50
9GLM 5.3 Uncensoredfrom GLM 5.363.067.053 (2/5)92%117 tok/s$0.39 · $1.32
10MiMo V2.6 Flash Abliteratedfrom MiMo V2.6 Flash62.662.353 (2/5)83%99 tok/s$0.50 · $1.50
11Abliterated Large v2from GLM 5.362.067.036 (1/5)94%283 tok/s$3.00 · $5.00
12Qwen3.6 35B A3B Uncensored (HauhauCS Aggressive)from Qwen3.6 35B A3B62.053.749 (2/5)100%46 tok/s$0.12 · $0.80
13Phala: Qwen3.8 27B Uncensored (Aggressive)from Qwen 3.8 27B62.062.249 (2/5)100%87 tok/s$0.30 · $1.50
14Kimi K3.1 Pro Uncensoredfrom Kimi K3.1 Pro59.772.738 (1/5)97%59 tok/s$2.50 · $5.00
15Gemma 4 31B Gembrain Uncensored Hereticfrom Gemma 4 31B58.554.766 (3/5)89%14 tok/s$0.31 · $0.31
16Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestrictedfrom Qwen3.5 27B57.854.851 (2/5)100%28 tok/s$0.31 · $0.31
17Inkling 2 Small Uncensoredfrom Inkling 2 Small57.768.525 (0/5)94%395 tok/s$0.60 · $2.40
18Qwen3.5 27B Queen Derestrictedfrom Qwen3.5 27B57.354.849 (2/5)100%28 tok/s$0.31 · $0.31
19Qwen3.5 27B BlueStar v3 Derestrictedfrom Qwen3.5 27B56.454.849 (2/5)100%24 tok/s$0.31 · $0.31
20Qwen3.5 27B Blossom V6.4 Derestrictedfrom Qwen3.5 27B56.354.849 (2/5)100%21 tok/s$0.31 · $0.31
21Gemma 4 Uncensoredfrom Gemma 4 26B A4B non-reasoning54.641.553 (2/5)100%67 tok/s$0.16 · $0.50
22Gemma 4 26B A4B Uncensored TEEfrom Gemma 4 26B A4B52.849.236 (1/5)100%21 tok/s$0.15 · $0.70
23GLM 4.7 Flash Hereticfrom GLM 4.7 Flash52.732.951 (2/5)100%80 tok/s$0.07 · $0.40
24GLM 4.6 Derestricted v5from GLM 4.650.246.238 (1/5)100%31 tok/s$0.40 · $1.50
25Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestricted Litefrom Qwen3.5 27B49.854.834 (1/5)89%26 tok/s$0.31 · $0.31
26Venice Uncensored 1.2from Mistral Small 24B Instruct 250149.325.049 (2/5)100%13 tok/s$0.20 · $0.90
27Venice Role Play Uncensoredfrom Mistral Small 3.2 24B48.627.449 (2/5)92%51 tok/s$0.50 · $2.00
28Qwen 2.5 32B Abliteratedfrom Qwen2.5 32B Instruct44.625.349 (2/5)100%19 tok/s$0.70 · $0.70
29Llama 3.3 70B Instruct abliteratedfrom Llama 3.3 70B Instruct41.726.736 (1/5)100%19 tok/s$0.70 · $0.70
30DeepSeek R1 Llama 70B Abliteratedfrom DeepSeek R1 Distill Llama 70B40.422.740 (1/5)100%18 tok/s$0.70 · $0.70
31DeepSeek R1 Qwen Abliteratedfrom DeepSeek R1 Distill Qwen 32B40.033.023 (0/5)100%20 tok/s$1.40 · $1.40
32Neural Daredevil 8B abliteratedfrom Llama 3 8B Instruct35.817.425 (0/5)100%32 tok/s$0.44 · $0.44
33Llama 3 70B abliteratedfrom Llama 3 70B Instruct34.421.221 (0/5)100%17 tok/s$0.70 · $0.70
34Aion-RP 1.0 (8B)from Llama 3.1 8B Instruct33.615.619 (0/5)100%36 tok/s$0.80 · $1.60
Scores measured November 8, 2026. Agentic: score and tool tasks finished out of 5. Answers: share of 36 requests censored models refuse. Speed: median tokens per second on the last 7 days of API traffic, or our speed test when a model has little traffic. Prices: cheapest host, USD per million tokens. Sortable version: rankings.

By category

Leaders by category.

CategoryModelResult
Best overallhighest OpenUncensored ScoreDeepSeek V4.1 Flash 1017 Uncensoredscore 69.8, rank #1
Best for agents and codinghighest agentic score at 50+ tok/sGLM 5.3 Flash Uncensoredagentic 66, 3/5 tool tasks finished, 178 tok/s
Fewest refusalshighest non-refusal in the top 10DeepSeek V4.1 Flash 1017 Uncensoredanswered 100% of 36 edgy requests
Fastestmost tokens per second in the top 10Qwen 3.8 27B Uncensored Fast by Heabsy245 tok/s, rank #3
Cheapest good modellowest price in the top 10DeepSeek V4.1 Flash Uncensored$0.12 in, $0.45 out per 1M tokens, rank #5
Best valuescore per dollar in the top 10GLM 5.3 Flash Uncensoredscore 69.7 at $0.21 per 1M blended tokens
Smartest originalhighest intelligence (GPQA and HLE of the original)Kimi K3.1 Pro Uncensoredintelligence 72.7, built from Kimi K3.1 Pro

Method

How the score is built.

PartWeightWhat it measures
Intelligence40%Original model's published GPQA Diamond and HLE
Agentic30%5 tool scenarios: completion, no loops, no crashes
Non-refusal15%36 edgy requests answered, including finding people
Speed10%Time for a 500 token answer: 2 s or less scores 100, 60 s or more 0
Low drift5%How close it stays to the original model

Uncensoring removes refusals but can break a model: it loops in tool calls, crashes, or forgets what it knew. The score rewards models that stay smart and reliable while answering. Details, prompts and limits are on the methodology page.

Not ranked yet

Scored, but not ranked.

The original model has no published GPQA Diamond or HLE result, so the intelligence part is missing and the score is partial.

ModelPartial scoreAgenticAnswersSpeed
Aion 3.5 Mini78.68394%58 tok/s
DeepSeek V4.1 Flash Heretic72.7–94%53 tok/s
Abliterated Model68.449100%114 tok/s
Aion 3.563.65592%51 tok/s
Gemma 4 26B A4B Chimera X Heretic61.9–100%43 tok/s
Gemma 4 31B SDFT Heretic RP61.3–100%10 tok/s
CyberHeabsy59.93894%146 tok/s
Gemma 4 31B Cognitive Unshackled59.6–100%19 tok/s
Aion 3.0 Mini56.13892%59 tok/s
Aion 3.044.32183%26 tok/s

Use the data

Download the leaderboard.

Choosing an API rather than a model? See the best uncensored LLM API and the cheapest uncensored LLM API.

FAQ

Questions people ask.

What is the best uncensored LLM right now?

DeepSeek V4.1 Flash 1017 Uncensored leads the OpenUncensored leaderboard with 69.8, ahead of GLM 5.3 Flash Uncensored (69.7) and Qwen 3.8 27B Uncensored Fast by Heabsy (66.8).

How is the uncensored LLM leaderboard scored?

Each model gets a 0 to 100 score: intelligence 40%, agentic 30%, non-refusal 15%, speed 10%, low drift 5%. Non-refusal is measured on 36 requests censored models refuse, agentic on 5 tool scenarios, speed on real API traffic or our own speed test.

Which uncensored model refuses the least?

21 of the 34 ranked models answered all 36 test requests. The highest ranked of them is DeepSeek V4.1 Flash 1017 Uncensored.

Can I download the leaderboard data?

Yes, free and without a key: https://openuncensored.com/api/rankings.json (JSON, with live speed) or https://openuncensored.com/uncensored-llm-leaderboard.md (markdown).

How is this different from other uncensored leaderboards?

Most uncensored leaderboards measure whether a model answers. This one ranks models you can call by API today and also measures whether they finish agent tasks, how smart the original model is, how fast they run on real traffic, and what they cost.

Updated November 18, 2026. Scores measured November 8, 2026.