GitHubGitHub RepoNeeds Review

Alishahryar1/free-claude-code

Free Claude Code is a local proxy that lets Claude Code, Codex, and Pi use cloud or local models through native pickers. It broadens model choice, but adds a credential-bearing gateway to audit and maintain.

GitHub

Open Source Open Markdown Open JSON

Source Summary

Free Claude Code routes Claude Code, Codex, and Pi through **31 cloud and local providers** from a local Admin UI. It preserves model selection, streaming, tools, reasoning controls, and image input where providers support them, with optional IDE and messaging integrations.

Practical Implication

Use it when you need provider portability or want to compare models without replacing your coding-agent clients. Prefer tool-capable models with sufficient context, protect the proxy with token authentication, and review the installer and generated client configuration before use.

Agent-Ready Context

Free Claude Code routes Claude Code, Codex, and Pi through **31 cloud and local providers** from a local Admin UI. It preserves model selection, streaming, tools, reasoning controls, and image input where providers support them, with optional IDE and messaging integrations.

Use it when you need provider portability or want to compare models without replacing your coding-agent clients. Prefer tool-capable models with sufficient context, protect the proxy with token authentication, and review the installer and generated client configuration before use.

Compatibility varies by provider, and unsupported reasoning controls fall back to provider behavior. The proxy stores credentials, changes endpoint configuration, and can expose agent sessions through Discord or Telegram, expanding the security and maintenance surface.

Connected Context · Feed7 Judgment

This moves model portability from a managed catalog into a local compatibility layer spanning three coding clients and many providers. It makes comparative testing and provider switching easier, but transfers credential custody, endpoint mutation, compatibility validation, and optional messaging exposure to the operator; breadth of access still does not identify a reliable default model.

Qwen 3.8 Max now available on Vercel AI GatewayQwen’s managed endpoint is one route this kind of comparison targets, while the proxy broadens evaluation across providers and clients at the cost of additional local security and maintenance duties.AI Gateway adds unified fast mode supportThe gateway standardizes a fast-mode request with fallback semantics, whereas this proxy must translate reasoning and performance controls across providers whose support varies.Kimi K3 and Kimi K3 Fast with ZDR and US-based providers now on AI GatewayKimi’s managed routes expose explicit speed, residency, retention, and fallback choices; the local proxy broadens provider choice but requires operators to verify those operational properties themselves.MiniMax M3 and M2.7 are free on AI GatewayThe MiniMax promotion is a bounded managed-gateway test with expiring IDs, while the proxy supports broader ongoing comparisons but introduces credential storage and configuration risk.

Context Map

toolscoding#coding-agents#model-selection#gateways

Uncertainty

Compatibility varies by provider, and unsupported reasoning controls fall back to provider behavior. The proxy stores credentials, changes endpoint configuration, and can expose agent sessions through Discord or Telegram, expanding the security and maintenance surface.