Method

Best unrestricted AI models, November 2026

The best unrestricted AI is Kimi K3.1 Pro, the largest model here with its refusals removed. 9 models are ranked below, with the size of each, how it lost its refusals and where it runs.

Updated November 18, 2026
Largest

Kimi K3.1 Pro (2,780B) is the largest unrestricted model, followed by DeepSeek V4.1 Flash 1017 (763B) and GLM-5.3 (753B).

Context window

Kimi K3.1 Pro, DeepSeek V4.1 Flash 1017, GLM-5.3 and GLM-5.3 Flash each hold 1,048,576 tokens.

Lowest price

DeepSeek V4.1 Flash 1017 has the lowest blended price, $0.41 per 1M tokens, followed by GLM-5.3 Flash at $0.43.

Images and video

Kimi K3.1 Pro reads text, images and video. Five more models here read images.

Runs locally

Llama 3.1 8B Abliterated runs on an 8 GB GPU. Qwen 3.6 35B-A3B Abliterated is the largest that fits one 24 GB GPU.

Leaderboard

Runs:
Refusals:
Refusals Size Context Features Price Availability
Model How refusals were removedParameterstotalContext windowtokensInputsIn / outUSD per 1MWhere it runsFurther analysis
#1Kimi K3.1 Pro Moonshot AI
AbliteratedWeights edited 2,780B 1M $3.50 / $8.00
Silk Compute
AbliteratedWeights edited 763B 1M $0.24 / $0.90
Silk Compute+2 more
#3GLM-5.3 Z.ai
AbliteratedWeights edited 753B 1M $2.00 / $4.00
Silk Compute+1 more
AbliteratedWeights edited 320B18B active 1M $0.24 / $1.00
Silk Compute
#5Hermes 3 405B Nous Research
Fine-tunedRetrained 405B 128K $1.20 / $1.20
Unfil.AI
AbliteratedWeights edited 36B3B active 256K Local only
Your machine24 GB of GPU memory
AbliteratedWeights edited 31.3B 256K Local only
Your machine24 GB of GPU memory
AbliteratedWeights edited 27.8B 256K $1.60 / $12.00
Featherlessor locally
AbliteratedWeights edited 8B 128K Local only
Your machine8 GB of GPU memory

Kimi K3.1 Pro leads, then the other models with a 1,048,576-token context, then the rest by total parameters. Prices are the lowest listed host's, read November 18, 2026. How we rank.

Size, context and price

Size

Total parameters, billions. Log scale. Larger is more capable

2,780 Kimi K3.1 Pro 763 DeepSeek V4.1 Flash 753 GLM-5.3 405 Hermes 3 405B 320 GLM-5.3 Flash 36 Qwen 3.6 35B 31.3 Gemma 4 31B 27.8 Qwen 3.8 27B 8 Llama 3.1 8B

Context window

Tokens the model holds at once. Higher is better

1M Kimi K3.1 Pro 1M DeepSeek V4.1 Flash 1M GLM-5.3 1M GLM-5.3 Flash 256K Qwen 3.6 35B 256K Gemma 4 31B 256K Qwen 3.8 27B 128K Hermes 3 405B 128K Llama 3.1 8B

Price

Blended USD per 1M tokens, lowest listed host. Lower is better

$0.41 DeepSeek V4.1 Flash $0.43 GLM-5.3 Flash $1.20 Hermes 3 405B $2.50 GLM-5.3 $4.20 Qwen 3.8 27B $4.63 Kimi K3.1 Pro
AbliteratedFine-tunedHatched: weights only, no API listed

Where to use them

Through an API

Seven providers serve unrestricted models. The largest is Kimi K3.1 Pro.

On a local GPU

Four models run on one consumer GPU or a Mac with one Ollama command.

In a chat app or agent

Any app that takes an OpenAI-compatible key can use these models.

Questions

What is the best unrestricted AI?

Kimi K3.1 Pro, at 2,780B parameters with a 1,048,576-token context. DeepSeek V4.1 Flash 1017 is the low-price pick at $0.24 in and $0.90 out.

Are there still limits?

Yes. Silk Compute's Terms forbid sexual content involving anyone under 18, and the law where you live still applies.

How is unrestricted different from uncensored?

In practice they mean the same thing. Jailbroken and filter-free AI are different.