claudish and agentic-flow
These are complements: claudish provides a polyglot Claude coding agent framework, while agentic-flow enables switching between alternative models within that agent framework to optimize cost and performance.
About claudish
MadAppGang/claudish
Claude Code. Any Model. The most powerful AI coding agent now speaks every language.
Implements an Anthropic API-compatible proxy server that routes Claude Code requests to 580+ models across cloud providers (OpenRouter, Gemini, OpenAI, OllamaCloud) and local runtimes (Ollama, LM Studio, vLLM). Leverages provider-specific routing syntax (`provider@model[:concurrency]`) with automatic model detection, enabling use of existing API subscriptions without additional fees. Includes vision support via automatic image description proxying for non-vision models and protocol compliance with Claude Code's communication layer.
About agentic-flow
ruvnet/agentic-flow
Easily switch between alternative low-cost AI models in Claude Code/Agent SDK. For those comfortable using Claude agents and commands, it lets you take what you've created and deploy fully hosted agents for real business purposes. Use Claude Code to get the agent working, then deploy it in your favorite cloud.
Combines SONA (Self-Optimizing Neural Architecture) with 213 MCP tools and 66 specialized agents using Flash Attention, GNN query refinement, and LoRA fine-tuning for <1ms adaptive learning and 60% cost savings through intelligent model routing. Integrates AgentDB@alpha for graph-based reasoning, multi-agent consensus via 5 attention mechanisms (Flash/Multi-Head/Linear/Hyperbolic/MoE), and background worker dispatch—enabling agents trained in Claude Code to autonomously coordinate and continually learn without catastrophic forgetting.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work