Contents
SolCard Inference is an AI service from SolCard: a chat app and an API, running open models, that keeps nothing you write. You pay for it up front, in crypto or from a SolCard card, and it answers the questions mainstream assistants turn down.
Two ways to use it
The chat app at app.solcardinference.ai works like any AI chat: ask, attach a file, turn on web search when you need current pages. Your chats are stored in your browser, on your device.
The API at https://api.solcardinference.ai/v1 speaks three formats, so the code and tools you already have work by changing two lines:
- OpenAI Chat Completions
- OpenAI Responses
- Anthropic Messages
from openai import OpenAI
client = OpenAI(base_url="https://api.solcardinference.ai/v1", api_key="sk-si-...")
Keys start with sk-si-. You create them on the dashboard’s API keys page, and each is shown once.
The models
| Model | What it’s for | Context |
|---|---|---|
solcard-inference-fast | The default: fastest and cheapest | ~1M tokens |
solcard-inference-large | The most capable; always reasons | 1M tokens |
solcard-inference-base | General purpose | 262k tokens |
They’re open-weight models with their refusal training removed. They take on the work mainstream assistants refuse:
- security research and malware analysis for defense;
- red-teaming, and hardening safety classifiers;
- synthetic training data.
They’re not for harming people or breaking the law. Fast, Large or Base helps you pick one.
Keeps nothing
Your messages and the answers are never logged or stored. Not on our servers, not in our logs, not at the model. We keep a billing line per request (time, key, model, tokens, cost), and that’s all.
Two switches change that, and only you turn them on:
- Search sends a query to a search engine, which keeps it by its own rules.
- Sync puts your chats on all your devices, end-to-end encrypted. Our servers hold a copy they can’t open.
Keeps nothing follows one message all the way through.
Paying for it
It’s prepaid. One USD balance pays for chat, every key and every model, and it never expires.
- Top up by sending crypto to your deposit address (SOL, USDC, ETH, BTC and more, on 9+ networks), or from a SolCard card balance.
- Pay as you go from $1.25 per million input tokens on Fast.
- Or take a plan: Starter $20, Pro $100 or Max $200 a month, each with more usage than the same money buys as you go.
New email sign-ups may get $1 of free credit for 30 days. How billing works has the details.
Signing up
Sign up with an email address or a Solana wallet. There’s no card on file.
In your coding agent
Paste this into Claude Code or Codex:
Add SolCard Inference to my models: solcardinference.ai/install
Our models show up next to your own. The guide covers what it changes.
Where to go next
- app.solcardinference.ai to start chatting
- docs.solcardinference.ai for the API
- Pricing for every rate