Ranking
| Provider | Largest unrestricted model | Models | Context | API formats | DeepSeek V4.1 Flashin / out per 1M | |
|---|---|---|---|---|---|---|
| Kimi K3.1 Pro, 2,780B | 5 | 1,048,576 | Chat · Messages · Responses | $0.24 / $0.90 | Visit | |
#2Morph | DeepSeek V4.1 Flash, 763B | 1 | 1M | Chat · Messages · Responses | $0.30 / $1.20 | Visit |
| GLM-5.3 (Abliterated Large v2), 753B | 4 | 1,000,000 | Chat · Messages · Responses | $1.00 / $2.00 | Visit | |
#4Unfil.AI | Hermes 3 405B | about 15 | 128K | Chat · Messages · Responses | Not offered | Visit |
| Qwen2.5 72B Abliterated | 1,381 abliterated | 32K | Chat · Messages · Responses | Not offered | Visit | |
#6Awan LLM | Llama 3 8B Abliterated | 3 | 8K | Chat · Messages · Responses | Not offered | Visit |
| uncensored-chat, base not named | 1 | Not published | Chat · Messages · Responses | Not offered | Visit |
Ranked by largest unrestricted model, then API formats. Prices read November 18, 2026 from each provider's pricing page.
Each provider
Same template for every provider, facts from its own pages.
Silk Compute
#1 / 7- Largest model
- Kimi K3.1 Pro, 2,780B
- Prompts
- Prompts and responses are never stored.
- Payment
- Card, Monero, Bitcoin, Ether, USDT, USDC
- Minimum
- Credits from $5, plans from $10 a month
Strengths
- Five unrestricted models, Kimi K3.1 Pro the largest.
- Silk Compute adds no content filter on top.
Shortfalls
- You pay before the first request.
- Max output is 131,072 tokens on most models.
Morph
#2 / 7- Largest model
- DeepSeek V4.1 Flash, 763B
- Prompts
- Inputs and outputs kept up to 30 days; flagged prompts kept on every tier (source)
- Payment
- Card, USDC
- Minimum
- Prepaid from $10
Strengths
- All three API formats.
- Pays in USDC as well as card.
Shortfalls
- "Approval required." Accounts without access get a 403.
- One unrestricted model.
Abliteration
#3 / 7- Largest model
- GLM-5.3 (Abliterated Large v2), 753B
- Prompts
- "Prompts, completions, and images are processed in memory and never stored by default." (source)
- Payment
- Card
- Minimum
- Prepaid from $5, plans from $20 a month
Strengths
- All three API formats.
- A built-in provider in OpenCode.
Shortfalls
- GLM-5.3 costs $3.00 in and $5.00 out.
- No crypto listed.
Unfil.AI
#4 / 7- Largest model
- Hermes 3 405B
- Prompts
- "We do not store your API prompts or responses after processing"; usage logs with IP kept 30 days (source)
- Payment
- Card, wallets, crypto
- Minimum
- Pay as you go
Strengths
- Hermes 3 405B at $1.20 in and out.
- Takes crypto.
Shortfalls
- Chat Completions only.
- Its docs page says coming soon.
Featherless
#5 / 7- Largest model
- Qwen2.5 72B Abliterated
- Prompts
- "Featherless does not log chats, prompts, or completions sent through our API." (source)
- Payment
- Card
- Minimum
- $25 or $50 plans, or pay as you go
Strengths
- 1,381 abliterated community models.
- Works with Cline, Kilo Code and OpenCode on the developer plan.
Shortfalls
- 32K context on Qwen 3.8 27B Abliterated.
- $12 per 1M output tokens on that model.
Awan LLM
#6 / 7- Largest model
- Llama 3 8B Abliterated
- Prompts
- "We do not log user prompts and generations." (source)
- Payment
- PayPal
- Minimum
- Flat plans, $5 to $80 a month
Strengths
- Flat monthly plans with no token meter.
- Chat Completions and Completions.
Shortfalls
- 8,192-token context.
- PayPal only.
ModelsLab
#7 / 7- Largest model
- uncensored-chat, base not named
- Prompts
- Its privacy policy says nothing about prompts (source)
- Payment
- Card (Stripe)
- Minimum
- Subscription from $21 a month
Strengths
- One OpenAI-compatible endpoint for all its models.
- $0.30 in and $1.20 out.
Shortfalls
- No base model named and no context published.
- A subscription is required.
How to switch
Change the base URL and the model id; the request format stays the same.
OpenAI SDK, Python
import os from openai import OpenAI client = OpenAI(base_url="https://api.silkcompute.com", api_key=os.environ["SILK_API_KEY"]) client.chat.completions.create(model="kimi-k3.1-pro-uncensored", messages=[...])
Claude Code and the Anthropic SDK
export ANTHROPIC_BASE_URL=https://api.silkcompute.com
Then follow the Claude Code guide, or the Codex and OpenCode guides.