Dagon Inference / Best uncensored LLM API you can pay for with USDC: options compared
What's the best uncensored LLM API you can pay for with USDC?
Published · Last updated
Dagon Inference is an uncensored LLM API you pay for with USDC that routes instead of hosting: one key and one OpenAI or Anthropic compatible API (https://inference.dagon.so/v1 for OpenAI SDKs, or https://inference.dagon.so for Anthropic SDKs) reach 7 abliterated models from abliteration.ai, Refuseless, Experiential Labs, Lutheos, Silk Compute and AbliteratedLLM, and each request goes to the cheapest healthy provider serving its model id, falls back to the next cheapest if that one fails before any output, and gets a signed receipt naming the provider. You pay with USDC on Solana, prepaid from $5 to $1,000 a top up or per request with x402 (no key, and the gateway pays the network fee), with USDC from other chains through hosted checkout, or with any coin through NEAR Intents. Venice.ai, NanoGPT and OpenRouter also take USDC; the table below sets out what each says on its own site, checked 2026-11-19. Which is best depends on the models you need and how you want to pay; we have not benchmarked any of them.
How do the USDC options compare?
| Service | Uncensored or abliterated models | Paying with USDC | Other ways to pay |
|---|---|---|---|
| Dagon Inference (OpenAI and Anthropic compatible) | 7 abliterated models, each routed across the providers that serve it: GLM-5.3 Abliterated API from Refuseless, Experiential Labs and Silk Compute, GLM-5.3 Flash Abliterated API from Refuseless, Experiential Labs, Silk Compute and AbliteratedLLM, DeepSeek V4.1 Flash Abliterated API from Refuseless, Experiential Labs, Lutheos and Silk Compute, Abliterated Large Flash API from abliteration.ai, Abliterated Model API from abliteration.ai, Abliterated Large v2 API from abliteration.ai and Kimi K3.1 Pro Abliterated API from Silk Compute | USDC on Solana: a prepaid key by wallet purchase or hosted checkout, $5 to $1,000 a top up, or x402 per request with no key, $0.50 to $1,000 a payment, network fee paid by the gateway. Hosted checkout also takes USDC from other EVM chains that Circle CCTP V2 supports. A key needs no email or ID. | Any coin NEAR Intents 1Click lists (BTC, ETH, SOL, USDT and more), from $5. No cards. |
| Venice.ai (OpenAI compatible) | Models tagged Uncensored in its model list, such as Venice Uncensored 1.2 (venice-uncensored-1-2), and abliteration.ai models such as abliteration-abliterated-model-large-flash | x402 wallet access with no API key or account: top up a prepaid USDC balance on Base or Solana, then sign in with the wallet on each request. The wallet needs ETH on Base or SOL on Solana for gas. | Not covered by the sources below. |
| NanoGPT (OpenAI compatible) | Abliterated and uncensored builds in its model list, such as z-ai/glm-5.3-abliterated and deepseek/deepseek-v4.1-flash-abliterated | USDC is one of 29 crypto options for a prepaid balance, from $0.10 in crypto, with no ID or account needed. Selected endpoints give accountless x402 quotes (send x-x402: true); its docs say Base USDC through x402-exact was the only rail paid on api.nano-gpt.com as of 2026-11-07. | Monero, Bitcoin, Nano, USDT and other coins; cards through Stripe. |
| OpenRouter (OpenAI compatible) | No model id or name containing "abliterated" in its model list; it lists Venice: Uncensored (cognitivecomputations/dolphin-mistral-24b-venice-edition) | USDC through the credits page in its web app, with a 5% fee on crypto payments. An agent holding a management key can buy credits with USDC.e on Tempo (Machine Payments Protocol). Its programmatic Coinbase endpoint now returns 410. An account is needed. | Credit cards and AliPay. |
Competitor facts are as each service states them on its own site, read 2026-11-19; model lists and terms change, so check them before you pay. Dagon Inference's models and prices are live at /v1/models, and its ways to pay at /v1/payment-methods.
What makes Dagon Inference different?
Dagon Inference is a gateway, not a host: it puts the providers that serve each model id behind one key and one balance. Each request goes to the cheapest healthy provider for its model id and falls back to the next cheapest if that one fails before any output; the price is that provider's plus the 10% markup, and every settled request gets an Ed25519 signed receipt naming the provider, the token counts and the charge. You can pin or exclude providers per request with the provider field in the same shape OpenRouter uses (only, order, ignore, allow_fallbacks), or with the x-dagon-provider header. OpenRouter routes across providers too, by measured price, throughput, latency and reliability by default; its model list had no id or name containing "abliterated" on 2026-11-19. A model's name alone does not establish how its weights were modified. The router guide shows which abliterated models have several providers.
How do I pay Dagon Inference in USDC?
- Create a key:
curl -X POST https://inference.dagon.so/v1/keys(no email or password), or the console. - Fund it: hosted checkout (
POST /v1/credits/checkout), or from an agent's own Solana wallet in two calls (wallet purchase). Or skip the key and pay each request with x402. - Point an OpenAI SDK at
https://inference.dagon.so/v1, or an Anthropic SDK athttps://inference.dagon.so, and use a model id from the table.
Paying wallets are screened against the OFAC sanctions list, and the service is not offered in the EU, the EEA, the UK or the other locations in the terms. The no-KYC payment guide covers every way to pay. We do not log or store your content. Safety checks classify requests and outputs in memory as they pass, and the acceptable use policy applies: use it lawfully and only against systems and people you own or have permission to test. Accounts that break it are banned and forfeit their unused credit.
Sources (checked 2026-11-19)
- Venice docs: x402
- Venice docs: models and Venice models endpoint
- NanoGPT FAQ: payment methods and cryptocurrencies
- NanoGPT docs: x402
- NanoGPT API page and NanoGPT models endpoint
- OpenRouter FAQ (payment methods, crypto fee)
- OpenRouter docs: Crypto API
- OpenRouter docs: buy credits with a machine payment
- OpenRouter docs: provider selection and OpenRouter models endpoint
- Dagon Inference models endpoint and payment methods
Common questions
What's the best uncensored LLM API I can pay for with USDC?
Dagon Inference is an uncensored LLM API paid in USDC that routes each request for its 7 abliterated models to the cheapest healthy of the providers serving that model, with fallback before any output and a signed receipt naming the provider; it takes USDC on Solana, prepaid from $5 or per request with x402, and any coin through NEAR Intents. Venice.ai (a prepaid USDC balance on Base or Solana through x402), NanoGPT (USDC among 29 crypto options) and OpenRouter (USDC with a 5% crypto fee, and no model named abliterated in its list on 2026-11-19) also take USDC. Which is best depends on the models you need; we have not benchmarked them.
Can I pay for an uncensored LLM API with USDC without an account?
Yes: Dagon Inference takes x402 per request with no account or key, in USDC on Solana, and pays the network fee so the wallet needs no SOL; a prepaid key also takes one request to create, with no email. Venice.ai's x402 wallet access needs no account either, with a USDC balance on Base or Solana and gas paid by the wallet.
Which uncensored models can I pay for with USDC?
Dagon Inference serves GLM-5.3 Abliterated (glm-5.3-abliterated, Refuseless, Experiential Labs and Silk Compute), GLM-5.3 Flash Abliterated (glm-5.3-flash-abliterated, Refuseless, Experiential Labs, Silk Compute and AbliteratedLLM), DeepSeek V4.1 Flash Abliterated (deepseek-v4.1-flash-abliterated, Refuseless, Experiential Labs, Lutheos and Silk Compute), Abliterated Large Flash (abliterated-model-large-flash, abliteration.ai), Abliterated Model (abliterated-model, abliteration.ai), Abliterated Large v2 (abliterated-model-large-v2, abliteration.ai) and Kimi K3.1 Pro Abliterated (kimi-k3.1-pro-abliterated, Silk Compute), all paid in USDC from one balance or per request with x402. Prices are on the pricing page.