Contents
OpenRouter and SolCard Inference solve different problems. OpenRouter is one key for hundreds of models from many providers. SolCard Inference is a small set of open models with their refusals removed, where nothing you write is kept. If you need GPT, Claude and Gemini behind one bill, use OpenRouter. If you need a model that won’t refuse and a provider that keeps nothing, read on.
Facts about OpenRouter below come from its own pages, as of October 31, 2026.
At a glance
| OpenRouter | SolCard Inference | |
|---|---|---|
| What it is | A router across many providers’ models | Three open models, refusal training removed |
| APIs | OpenAI-compatible; Anthropic-compatible | OpenAI Chat Completions and Responses; Anthropic Messages |
| Token prices | The providers’ prices, no markup | Our own price list |
| Fees | 5.5% when you buy credits (Standard) | None on top of the price list |
| Plans | Pay as you go | Pay as you go, or $20 / $100 / $200 a month |
| Pay with | Cards, AliPay, USDC | Crypto on 9+ networks, or a SolCard card balance |
| Prompts kept | Not by OpenRouter by default; depends on the provider each request reaches | Never |
| Chat app | Yes | Yes, history on your device |
Privacy: per request, or by default
OpenRouter doesn’t store your prompts by default. Its data collection page says it “does not store your prompts or responses” unless you opt in to logging. But the request then goes to a model provider, and its privacy policy is plain that “some Model Providers may use your Inputs and Outputs for model training or improvement.”
You can close that gap. OpenRouter’s zero data retention setting routes requests “only to endpoints that have a Zero Data Retention policy,” per request or for your whole account. It’s a setting you have to know about and turn on, and it narrows which providers you can reach.
At SolCard Inference, zero retention isn’t a setting. Every request, on every model, is processed in memory and not stored: not on our servers, not in our logs, not at the model. Your chat history stays in your browser unless you turn on end-to-end encrypted sync. Keeps nothing follows one message all the way through.
Models: hundreds, or three that answer
OpenRouter’s strength is choice: the big closed models and most open ones, behind one key. Its terms let it screen and block inputs, and each model keeps its own refusals.
SolCard Inference has three models, open-weight with their refusal training removed: Fast (the default), Large (always reasons) and Base. They take on what mainstream models refuse, like security research, malware analysis for defense, red-teaming and synthetic training data. Fast, Large or Base compares them.
Price: pass-through plus a fee, or one price list
OpenRouter passes through each provider’s token price, with “no markup on inference pricing” (FAQ). It charges when you buy credits instead: “5.5% on Standard and 8% on Business” (pricing). Purchases run from $5 to $25,000, and its terms let it expire unused credits after 365 days.
SolCard Inference has one price list, with no fee when you top up. Per million tokens on Fast: $1.25 input, $0.125 cached input, $2.50 output. Your balance never expires. Plans (Starter $20, Pro $100, Max $200 a month) buy 20% to 25% more usage than the same money as you go. How billing works has the details.
Paying in crypto
OpenRouter takes “all major credit cards, AliPay and cryptocurrency payments in USDC” (FAQ). Its terms say “cryptocurrency payments are never refundable.”
SolCard Inference is built for crypto: send SOL, USDC, ETH, BTC and more, on 9+ networks, to your deposit address, or pay from a SolCard card balance. There’s no card on file. Sign up with an email address or a Solana wallet.
Coding agents
OpenRouter documents Claude Code through its Anthropic-compatible endpoint, which replaces Claude Code’s connection to Anthropic. It notes the setup “is only guaranteed to work with the Anthropic first-party provider.”
SolCard Inference adds its models next to your own in Claude Code and Codex. Paste Add SolCard Inference to my models: solcardinference.ai/install and your Claude or ChatGPT models stay on your own plan. The guide shows what it changes.
Which one
- Use OpenRouter for the widest choice of models behind one key, especially closed ones like GPT, Claude and Gemini.
- Use SolCard Inference when you want models that don’t refuse, zero retention on every request without a setting, and a prepaid balance you top up in crypto.
You can use both: a router for the closed models, and SolCard Inference for the work they won’t do.