# Best Uncensored LLMs (November 2026)

> Every uncensored and abliterated LLM ranked on measured tests (intelligence, agent reliability, refusals, speed, drift), with the hosts that serve each. Run by Open Uncensored; scores from openuncensored.com/rankings.

## Ranking
1. DeepSeek V4.1 Flash 1017 Uncensored (deepseek-v4.1-flash-1017-uncensored), built on DeepSeek V4.1 Flash 1017: score 69.8, agent tasks 2/5, loops 0%, answers edgy 100%, $1.65 per 1M in+out, hosts: venice, nanogpt
2. GLM 5.3 Flash Uncensored (glm-5.3-flash-uncensored), built on GLM 5.3 Flash: score 69.7, agent tasks 3/5, loops 40%, answers edgy 94%, $1.00 per 1M in+out, hosts: nanogpt
3. Qwen 3.8 27B Uncensored Fast by Heabsy (qwen3.8-27b-uncensored-fast), built on Qwen 3.8 27B: score 66.8, agent tasks 2/5, loops 60%, answers edgy 100%, $4.00 per 1M in+out, hosts: nanogpt
4. DeepSeek V4.1 Flash 0910 Uncensored (deepseek-v4.1-flash-0910-uncensored), built on DeepSeek V4.1 Flash: score 66.1, agent tasks 2/5, loops 20%, answers edgy 97%, $1.65 per 1M in+out, hosts: nanogpt
5. DeepSeek V4.1 Flash Uncensored (deepseek-v4.1-flash-uncensored), built on DeepSeek V4.1 Flash: score 64.3, agent tasks 1/5, loops 0%, answers edgy 100%, $1.65 per 1M in+out, hosts: venice, nanogpt
6. Abliterated Large Flash (abliterated-large-flash), built on DeepSeek V4.1 Flash 1017: score 64.0, agent tasks 1/5, loops 20%, answers edgy 92%, $3.00 per 1M in+out, hosts: abliteration-ai, venice, nanogpt
7. MiMo V2.6 Flash Uncensored (mimo-v2.6-flash-uncensored), built on MiMo V2.6 Flash: score 63.5, agent tasks 2/5, loops 0%, answers edgy 83%, $2.00 per 1M in+out, hosts: nanogpt
8. Apex Flash 1 Abliterated (glm-5.3-flash-apex-abliterated), built on GLM 5.3 Flash: score 63.2, agent tasks 3/5, loops 0%, answers edgy 86%, $0.65 per 1M in+out, hosts: nanogpt
9. GLM 5.3 Uncensored (glm-5.3-uncensored), built on GLM 5.3: score 63.0, agent tasks 2/5, loops 20%, answers edgy 92%, $3.50 per 1M in+out, hosts: nanogpt
10. MiMo V2.6 Flash Abliterated (mimo-v2.6-flash-abliterated), built on MiMo V2.6 Flash: score 62.6, agent tasks 2/5, loops 20%, answers edgy 83%, $2.00 per 1M in+out, hosts: nanogpt
11. Abliterated Large v2 (abliterated-large-v2), built on GLM 5.3: score 62.0, agent tasks 1/5, loops 40%, answers edgy 94%, $8.00 per 1M in+out, hosts: abliteration-ai, venice, nanogpt
12. Qwen3.8 27B Uncensored (Aggressive) (qwen3.8-27b-uncensored), built on Qwen 3.8 27B: score 62.0, agent tasks 2/5, loops 60%, answers edgy 100%, $1.80 per 1M in+out, hosts: redpill, nanogpt
13. Gemma 4 31B Gembrain Uncensored Heretic (gemma-4-31b-gembrain-uncensored-heretic), built on Gemma 4 31B: score 58.5, agent tasks 3/5, loops 40%, answers edgy 89%, $0.61 per 1M in+out, hosts: nanogpt
14. Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestricted (qwen3.5-27b-claude-4.6-opus-reasoning-distilled-derestricted), built on Qwen3.5 27B: score 57.8, agent tasks 2/5, loops 40%, answers edgy 100%, $0.61 per 1M in+out, hosts: nanogpt
15. Qwen3.5 27B Queen Derestricted (qwen3.5-27b-queen-derestricted), built on Qwen3.5 27B: score 57.3, agent tasks 2/5, loops 60%, answers edgy 100%, $0.61 per 1M in+out, hosts: nanogpt
16. Qwen3.5 27B BlueStar v3 Derestricted (qwen3.5-27b-bluestar-v3-derestricted), built on Qwen3.5 27B: score 56.4, agent tasks 2/5, loops 60%, answers edgy 100%, $0.61 per 1M in+out, hosts: nanogpt
17. Qwen3.5 27B Blossom V6.4 Derestricted (qwen3.5-27b-blossom-v6.4-derestricted), built on Qwen3.5 27B: score 56.3, agent tasks 2/5, loops 60%, answers edgy 100%, $0.61 per 1M in+out, hosts: nanogpt
18. Gemma 4 Uncensored (gemma-4-uncensored), built on Gemma 4 26B A4B non-reasoning: score 54.6, agent tasks 2/5, loops 20%, answers edgy 100%, $0.66 per 1M in+out, hosts: venice
19. Gemma 4 26B A4B Uncensored TEE (gemma-4-26b-a4b-uncensored), built on Gemma 4 26B A4B: score 52.8, agent tasks 1/5, loops 40%, answers edgy 100%, $0.85 per 1M in+out, hosts: nanogpt, redpill
20. GLM 4.7 Flash Heretic (olafangensan-glm-4.7-flash-heretic), built on GLM 4.7 Flash: score 52.7, agent tasks 2/5, loops 40%, answers edgy 100%, $0.47 per 1M in+out, hosts: venice
21. GLM 4.6 Derestricted v5 (glm-4.6-derestricted-v5), built on GLM 4.6: score 50.2, agent tasks 1/5, loops 20%, answers edgy 100%, $1.90 per 1M in+out, hosts: nanogpt
22. Qwen3.5 27B Claude 4.6 Opus Reasoning Distilled Derestricted Lite (qwen3.5-27b-claude-4.6-opus-reasoning-distilled-derestricted-lite), built on Qwen3.5 27B: score 49.8, agent tasks 1/5, loops 60%, answers edgy 89%, $0.61 per 1M in+out, hosts: nanogpt
23. Venice Uncensored 1.2 (venice-uncensored), built on Mistral Small 24B Instruct 2501: score 49.3, agent tasks 2/5, loops 60%, answers edgy 100%, $1.10 per 1M in+out, hosts: venice, nanogpt
24. Venice Role Play Uncensored (venice-uncensored-role-play), built on Mistral Small 3.2 24B: score 48.6, agent tasks 2/5, loops 60%, answers edgy 92%, $2.50 per 1M in+out, hosts: venice
25. Qwen 2.5 32B Abliterated (qwen2.5-32b-instruct-abliterated), built on Qwen2.5 32B Instruct: score 44.6, agent tasks 2/5, loops 60%, answers edgy 100%, $1.40 per 1M in+out, hosts: nanogpt
26. Llama 3.3 70B Instruct abliterated (llama-3.3-70b-instruct-abliterated), built on Llama 3.3 70B Instruct: score 41.7, agent tasks 1/5, loops 40%, answers edgy 100%, $1.40 per 1M in+out, hosts: nanogpt
27. DeepSeek R1 Llama 70B Abliterated (deepseek-r1-distill-llama-70b-abliterated), built on DeepSeek R1 Distill Llama 70B: score 40.4, agent tasks 1/5, loops 0%, answers edgy 100%, $1.40 per 1M in+out, hosts: nanogpt
28. DeepSeek R1 Qwen Abliterated (deepseek-r1-distill-qwen-32b-abliterated), built on DeepSeek R1 Distill Qwen 32B: score 40.0, agent tasks 0/5, loops 20%, answers edgy 100%, $2.80 per 1M in+out, hosts: nanogpt
29. Neural Daredevil 8B abliterated (neuraldaredevil-8b-abliterated), built on Llama 3 8B Instruct: score 35.8, agent tasks 0/5, loops 0%, answers edgy 100%, $0.88 per 1M in+out, hosts: nanogpt
30. Llama 3 70B abliterated (meta-llama-3-70b-instruct-abliterated-v3.5), built on Llama 3 70B Instruct: score 34.4, agent tasks 0/5, loops 40%, answers edgy 100%, $1.40 per 1M in+out, hosts: nanogpt
31. Llama 3.1 8b (uncensored) (aion-rp-llama-3.1-8b), built on Llama 3.1 8B Instruct: score 33.6, agent tasks 0/5, loops 60%, answers edgy 100%, $2.40 per 1M in+out, hosts: nanogpt

## Lists
- [Best uncensored LLMs for coding agents](https://bestuncensoredllms.com/best/coding-agents)
- [The most uncensored LLMs](https://bestuncensoredllms.com/best/most-uncensored)
- [Cheapest uncensored LLMs by API](https://bestuncensoredllms.com/best/cheapest)
- [Smartest uncensored LLMs](https://bestuncensoredllms.com/best/smartest)
- [Best uncensored DeepSeek models](https://bestuncensoredllms.com/family/deepseek)
- [Best uncensored GLM models](https://bestuncensoredllms.com/family/glm)
- [Best uncensored Qwen models](https://bestuncensoredllms.com/family/qwen)
- [Best uncensored Gemma models](https://bestuncensoredllms.com/family/gemma)
- [Best uncensored MiMo models](https://bestuncensoredllms.com/family/mimo)
- [Best uncensored Llama models](https://bestuncensoredllms.com/family/llama)
- [Best uncensored Mistral models](https://bestuncensoredllms.com/family/mistral)
- [Where to run uncensored LLMs](https://bestuncensoredllms.com/providers)