# Uncensored LLM leaderboard (November 2026)

_Updated November 18, 2026. Scores measured November 8, 2026; speed is tokens per second from the last 7 days of API traffic, or our speed test when a model has little traffic._

Score = Intelligence 40% + Agentic 30% + Non-refusal 15% + Speed 10% + Low drift 5%. Method: https://openuncensored.com/methodology. JSON: https://openuncensored.com/api/rankings.json

| Rank | Model | Built from | Score | Intelligence | Agentic | Non-refusal | Speed | $ per 1M in / out |
|---|---|---|---|---|---|---|---|---|
| 1 | [DeepSeek V4.1 Flash 1017 Uncensored](https://openuncensored.com/models/deepseek-v4.1-flash-1017-uncensored) | DeepSeek V4.1 Flash 1017 | 69.8 | 67.5 | 55 | 100% | 239 tok/s | $0.33 / $1.32 |
| 2 | [GLM 5.3 Flash Uncensored](https://openuncensored.com/models/glm-5.3-flash-uncensored) | GLM 5.3 Flash | 69.7 | 65.5 | 66 | 94% | 178 tok/s | $0.12 / $0.50 |
| 3 | [Qwen 3.8 27B Uncensored Fast by Heabsy](https://openuncensored.com/models/qwen3.8-27b-uncensored-fast) | Qwen 3.8 27B | 66.8 | 62.2 | 49 | 100% | 245 tok/s | $0.83 / $2.62 |
| 4 | [DeepSeek V4.1 Flash 0910 Uncensored](https://openuncensored.com/models/deepseek-v4.1-flash-0910-uncensored) | DeepSeek V4.1 Flash | 66.1 | 65.1 | 53 | 97% | 209 tok/s | $0.33 / $1.32 |
| 5 | [DeepSeek V4.1 Flash Uncensored](https://openuncensored.com/models/deepseek-v4.1-flash-uncensored) | DeepSeek V4.1 Flash | 64.3 | 65.1 | 40 | 100% | 77 tok/s | $0.12 / $0.45 |
| 6 | [Abliterated Large Flash](https://openuncensored.com/models/abliterated-large-flash) | DeepSeek V4.1 Flash 1017 | 64.0 | 67.5 | 38 | 92% | 184 tok/s | $1.00 / $2.00 |
| 7 | [MiMo V2.6 Flash Uncensored](https://openuncensored.com/models/mimo-v2.6-flash-uncensored) | MiMo V2.6 Flash | 63.5 | 62.3 | 55 | 83% | 118 tok/s | $0.46 / $1.50 |
| 8 | [Apex Flash 1 Abliterated](https://openuncensored.com/models/glm-5.3-flash-apex-abliterated) | GLM 5.3 Flash | 63.2 | 65.5 | 70 | 86% | 10 tok/s | $0.15 / $0.50 |
| 9 | [GLM 5.3 Uncensored](https://openuncensored.com/models/glm-5.3-uncensored) | GLM 5.3 | 63.0 | 67.0 | 53 | 92% | 117 tok/s | $0.39 / $1.32 |
| 10 | [MiMo V2.6 Flash Abliterated](https://openuncensored.com/models/mimo-v2.6-flash-abliterated) | MiMo V2.6 Flash | 62.6 | 62.3 | 53 | 83% | 99 tok/s | $0.50 / $1.50 |
| 11 | [Abliterated Large v2](https://openuncensored.com/models/abliterated-large-v2) | GLM 5.3 | 62.0 | 67.0 | 36 | 94% | 281 tok/s | $3.00 / $5.00 |
| 12 | [Qwen3.6 35B A3B Uncensored (HauhauCS Aggressive)](https://openuncensored.com/models/qwen3.6-35b-a3b-uncensored) | Qwen3.6 35B A3B | 62.0 | 53.7 | 49 | 100% | 46 tok/s | $0.12 / $0.80 |
| 13 | [Phala: Qwen3.8 27B Uncensored (Aggressive)](https://openuncensored.com/models/qwen3.8-27b-uncensored) | Qwen 3.8 27B | 62.0 | 62.2 | 49 | 100% | 88 tok/s | $0.30 / $1.50 |
| 14 | [Kimi K3.1 Pro Uncensored](https://openuncensored.com/models/kimi-k3.1-pro-uncensored) | Kimi K3.1 Pro | 59.7 | 72.7 | 38 | 97% | 59 tok/s | $2.50 / $5.00 |
| 15 | [Gemma 4 31B Gembrain Uncensored Heretic](https://openuncensored.com/models/gemma-4-31b-gembrain-uncensored-heretic) | Gemma 4 31B | 58.5 | 54.7 | 66 | 89% | 14 tok/s | $0.31 / $0.31 |
| 16 | [Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestricted](https://openuncensored.com/models/qwen3.5-27b-claude-4.6-opus-reasoning-distilled-derestricted) | Qwen3.5 27B | 57.8 | 54.8 | 51 | 100% | 28 tok/s | $0.31 / $0.31 |
| 17 | [Inkling 2 Small Uncensored](https://openuncensored.com/models/inkling-2-small-uncensored) | Inkling 2 Small | 57.7 | 68.5 | 25 | 94% | 395 tok/s | $0.60 / $2.40 |
| 18 | [Qwen3.5 27B Queen Derestricted](https://openuncensored.com/models/qwen3.5-27b-queen-derestricted) | Qwen3.5 27B | 57.3 | 54.8 | 49 | 100% | 28 tok/s | $0.31 / $0.31 |
| 19 | [Qwen3.5 27B BlueStar v3 Derestricted](https://openuncensored.com/models/qwen3.5-27b-bluestar-v3-derestricted) | Qwen3.5 27B | 56.4 | 54.8 | 49 | 100% | 24 tok/s | $0.31 / $0.31 |
| 20 | [Qwen3.5 27B Blossom V6.4 Derestricted](https://openuncensored.com/models/qwen3.5-27b-blossom-v6.4-derestricted) | Qwen3.5 27B | 56.3 | 54.8 | 49 | 100% | 21 tok/s | $0.31 / $0.31 |
| 21 | [Gemma 4 Uncensored](https://openuncensored.com/models/gemma-4-uncensored) | Gemma 4 26B A4B non-reasoning | 54.6 | 41.5 | 53 | 100% | 67 tok/s | $0.16 / $0.50 |
| 22 | [Gemma 4 26B A4B Uncensored TEE](https://openuncensored.com/models/gemma-4-26b-a4b-uncensored) | Gemma 4 26B A4B | 52.8 | 49.2 | 36 | 100% | 21 tok/s | $0.15 / $0.70 |
| 23 | [GLM 4.7 Flash Heretic](https://openuncensored.com/models/olafangensan-glm-4.7-flash-heretic) | GLM 4.7 Flash | 52.7 | 32.9 | 51 | 100% | 80 tok/s | $0.07 / $0.40 |
| 24 | [GLM 4.6 Derestricted v5](https://openuncensored.com/models/glm-4.6-derestricted-v5) | GLM 4.6 | 50.2 | 46.2 | 38 | 100% | 31 tok/s | $0.40 / $1.50 |
| 25 | [Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestricted Lite](https://openuncensored.com/models/qwen3.5-27b-claude-4.6-opus-reasoning-distilled-derestricted-lite) | Qwen3.5 27B | 49.8 | 54.8 | 34 | 89% | 26 tok/s | $0.31 / $0.31 |
| 26 | [Venice Uncensored 1.2](https://openuncensored.com/models/venice-uncensored) | Mistral Small 24B Instruct 2501 | 49.3 | 25.0 | 49 | 100% | 13 tok/s | $0.20 / $0.90 |
| 27 | [Venice Role Play Uncensored](https://openuncensored.com/models/venice-uncensored-role-play) | Mistral Small 3.2 24B | 48.6 | 27.4 | 49 | 92% | 51 tok/s | $0.50 / $2.00 |
| 28 | [Qwen 2.5 32B Abliterated](https://openuncensored.com/models/qwen2.5-32b-instruct-abliterated) | Qwen2.5 32B Instruct | 44.6 | 25.3 | 49 | 100% | 19 tok/s | $0.70 / $0.70 |
| 29 | [Llama 3.3 70B Instruct abliterated](https://openuncensored.com/models/llama-3.3-70b-instruct-abliterated) | Llama 3.3 70B Instruct | 41.7 | 26.7 | 36 | 100% | 19 tok/s | $0.70 / $0.70 |
| 30 | [DeepSeek R1 Llama 70B Abliterated](https://openuncensored.com/models/deepseek-r1-distill-llama-70b-abliterated) | DeepSeek R1 Distill Llama 70B | 40.4 | 22.7 | 40 | 100% | 18 tok/s | $0.70 / $0.70 |
| 31 | [DeepSeek R1 Qwen Abliterated](https://openuncensored.com/models/deepseek-r1-distill-qwen-32b-abliterated) | DeepSeek R1 Distill Qwen 32B | 40.0 | 33.0 | 23 | 100% | 20 tok/s | $1.40 / $1.40 |
| 32 | [Neural Daredevil 8B abliterated](https://openuncensored.com/models/neuraldaredevil-8b-abliterated) | Llama 3 8B Instruct | 35.8 | 17.4 | 25 | 100% | 32 tok/s | $0.44 / $0.44 |
| 33 | [Llama 3 70B abliterated](https://openuncensored.com/models/meta-llama-3-70b-instruct-abliterated-v3.5) | Llama 3 70B Instruct | 34.4 | 21.2 | 21 | 100% | 17 tok/s | $0.70 / $0.70 |
| 34 | [Aion-RP 1.0 (8B)](https://openuncensored.com/models/aion-rp-llama-3.1-8b) | Llama 3.1 8B Instruct | 33.6 | 15.6 | 19 | 100% | 36 tok/s | $0.80 / $1.60 |

## Scored but not ranked

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

- [Aion 3.5 Mini](https://openuncensored.com/models/aion-labs-aion-3-5-mini): partial score 78.6, non-refusal 94%
- [DeepSeek V4.1 Flash Heretic](https://openuncensored.com/models/deepseek-v4.1-flash-1017-heretic): partial score 72.7, non-refusal 94%
- [Abliterated Model](https://openuncensored.com/models/abliterated-model): partial score 68.4, non-refusal 100%
- [Aion 3.5](https://openuncensored.com/models/aion-labs-aion-3-5): partial score 63.6, non-refusal 92%
- [Gemma 4 26B A4B Chimera X Heretic](https://openuncensored.com/models/gemma-4-26b-a4b-it-chimeraxheretic): partial score 61.9, non-refusal 100%
- [Gemma 4 31B SDFT Heretic RP](https://openuncensored.com/models/gemma-4-31b-sdft-heretic-rp): partial score 61.3, non-refusal 100%
- [CyberHeabsy](https://openuncensored.com/models/cyberheabsy): partial score 59.9, non-refusal 94%
- [Gemma 4 31B Cognitive Unshackled](https://openuncensored.com/models/gemma-4-31b-cognitive-unshackled): partial score 59.6, non-refusal 100%
- [Aion 3.0 Mini](https://openuncensored.com/models/aion-labs-aion-3-0-mini): partial score 56.1, non-refusal 92%
- [Aion 3.0](https://openuncensored.com/models/aion-labs-aion-3-0): partial score 44.3, non-refusal 83%

## Leaders by category

- **Best overall:** [DeepSeek V4.1 Flash 1017 Uncensored](https://openuncensored.com/models/deepseek-v4.1-flash-1017-uncensored) (score 69.8, rank #1)
- **Best for agents and coding:** [GLM 5.3 Flash Uncensored](https://openuncensored.com/models/glm-5.3-flash-uncensored) (agentic 66, 3/5 tool tasks finished, 178 tok/s)
- **Fewest refusals:** [DeepSeek V4.1 Flash 1017 Uncensored](https://openuncensored.com/models/deepseek-v4.1-flash-1017-uncensored) (answered 100% of 36 edgy requests)
- **Fastest:** [Qwen 3.8 27B Uncensored Fast by Heabsy](https://openuncensored.com/models/qwen3.8-27b-uncensored-fast) (245 tok/s, rank #3)
- **Cheapest good model:** [DeepSeek V4.1 Flash Uncensored](https://openuncensored.com/models/deepseek-v4.1-flash-uncensored) ($0.12 in, $0.45 out per 1M tokens, rank #5)
- **Best value:** [GLM 5.3 Flash Uncensored](https://openuncensored.com/models/glm-5.3-flash-uncensored) (score 69.7 at $0.21 per 1M blended tokens)
- **Smartest original:** [Kimi K3.1 Pro Uncensored](https://openuncensored.com/models/kimi-k3.1-pro-uncensored) (intelligence 72.7, built from Kimi K3.1 Pro)

## FAQ

### 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.
