Integrate once. Access many models.
Compatible with OpenAI and Anthropic APIs — change your base_url to access Claude, OpenAI GPT, Grok, and other leading AI models, without changing your request format.
Govern AI usage across calls, keys, and workspaces.
Token-level pricing + workspace-level budget control + multi-layer cost breakdown — your enterprise AI spend, managed in one place.
Token-level pricing
Priced in separate buckets for input, output and cache — every charge traceable and reconcilable line by line.
Workspace-level budget control
Budget caps per workspace, spend caps per key — over-limit calls are blocked on the spot.
Multi-layer cost breakdown
Break spend down by organization / workspace / API key — allocation at a glance, every entry reconcilable.
Monitor usage in ModelGo, forward logs to your existing stack.
Complete usage dashboards and call logs — spend / requests / tokens tri-view, trends and distributions at a glance. Filter by model, status, key, and inspect raw payloads. Forward to Datadog, OpenTelemetry, and more.
Permission system that scales with your team
Manage members, workspaces, roles, and access from one enterprise account. Add people, delegate permissions, spin up projects — no rebuild required.
Every call stays within allowed boundaries
Safety guardrails cover model blocklists, IP allowlists, budget controls, and prompt-injection detection. Apply workspace defaults, then inherit or configure handling at the API key level.
Model blocklist
Restrict which models can be called — unauthorized ones are blocked outright.
IP allowlist
Restrict call sources — only trusted addresses get through.
Key-bound policies
Inherit workspace defaults or configure model, IP, budget, and prompt-injection handling for an API key.
Standard protocol, no project rework needed
OpenAI / Anthropic compatible — swap base_url and key, done. cURL / Python / Node examples included, first call in minutes.
Configure your AI coding tools from one CLI.
One command wires ModelGo into Claude Code, Codex and 12+ major AI coding clients — paste your key and go, with config written and connectivity verified automatically. Use -r to switch region.
$ npx @model-go/config -r global$ npx @model-go/config -k sk-xxx --only claude,codex$ npx @model-go/config -k sk-xxx --only all