Client compatibility
Connect coding tools through API-compatible endpoints
CodeUnlim exposes API formats used by a range of developer clients. Compatibility depends on each client’s custom-provider options and supported protocol features.
Available protocol surfaces
- OpenAI-compatible Chat Completions:
/v1/chat/completions. - Anthropic-compatible Messages:
/v1/messages. - Model ID:
deepseek-v4-flash. - Maximum context window: 262,144 tokens. Input and generated output share this window.
- Gateway output-token request ceiling: 65,536 tokens; available output can be lower depending on input size.
Requests require a CodeUnlim API key. Keep the key private, store it in the client’s supported secret store, and never place it in source control or browser analytics.
OpenCode and other custom-provider clients
OpenCode documents a custom provider configuration for OpenAI-compatible APIs. Configure the CodeUnlim endpoint, credential, and model using the current OpenCode provider instructions. The customer dashboard contains account-specific connection details.
We do not claim an official OpenCode partnership or guarantee that every optional tool feature works through every client. Check client protocol support and test a harmless request before using a workflow.
Claude Code and Anthropic-compatible clients
The gateway has an Anthropic Messages-compatible route. Whether a client can target a custom endpoint depends on that client’s own configuration options and terms. CodeUnlim is not affiliated with Anthropic, and this page does not promise feature parity with Anthropic’s hosted service.
Limits and safe use
Package stream counts define reserved access entitlement, not unlimited concurrent GPU decoding. Requests may queue when the per-package generation limit or runtime admission capacity is occupied. Model behavior, availability, supported request fields, and performance can change with deployment conditions.
Coming Soon or read the Terms of Service.