Add BlockRun to Dify as an OpenAI-API-compatible model provider
Set Dify's OpenAI-API-compatible provider to BlockRun's API-key endpoint and every app, agent and workflow in the workspace can pick any catalog model, on one invoice.
How it works
Dify is a hosted or self-hosted app builder, and its model settings include an OpenAI-API-compatible provider that takes a base URL, a key and a model name. Configured against BlockRun's API host with a key issued to your account, each catalog model becomes a model the workspace's apps, agents and workflows can select, metered at exact usage on one monthly invoice.
That is the invoiced door, which is the right one for a workspace many people share: one key for the team, one bill in dollars, no per-vendor accounts. A self-hosted Dify next to a blockrun-litellm sidecar can use the pay-per-call door instead with the same provider settings.
Behind it is the BlockRun catalog: 78 chat models plus image, video, music and speech generation, search, market data and multi-chain RPC, priced per call. 6 open-weight models are free and need no wallet.
Setup
1. Settings → Model Provider → OpenAI-API-compatible
API endpoint URL: https://api.blockrun.ai/v1
API Key: <your BlockRun key>
Model Name: anthropic/claude-sonnet-5 (add one per catalog model you want)Questions
- How do I add BlockRun to Dify?
- Add an OpenAI-API-compatible model provider with BlockRun's API host as the endpoint URL, your key, and a catalog model id as the model name; repeat for each model you want in the picker.
- Does Dify Cloud work with BlockRun?
- Yes, through the API-key door. Self-hosted Dify can also use a local sidecar on the pay-per-call door.
- Which models can Dify apps use through BlockRun?
- Any chat model on the catalog. Each one you add to the provider appears in the model selector for apps, agents and workflows.
- How is usage billed for a Dify workspace?
- At exactly what each call used, in dollars, on a monthly invoice, with no per-call minimum — the ledger is the bill.
- Can Dify's agent tools call BlockRun's search or image endpoints?
- Yes, as custom tools defined from the OpenAPI spec with the same key; the docs carry the endpoints.
- Dify on BlockRun vs Dify with a key per provider?
- One provider entry and one invoice for every vendor's models, and a new model is one more model name rather than a new credential and a new bill.
- What does a call cost through this integration?
- The provider's published token rate with no markup on tokens, plus a flat per-call fee, quoted in dollars in the 402 before the call runs. Image, video and speech calls are priced per unit the same way. The live rate card is on the models page.