if you're already bouncing between GPT, Claude and Grok, maintaining separate model integrations gets old fast that's what we're building Routera for: one model layer GPT + Claude + Grok Claude Code + Codex + OpenCode + Hermes usage-based access + lower-cost routing for