codaicodai

One endpoint · every frontier model

One model name. Every frontier model.

Point your tools at codai. The router picks the right frontier model for each task, checks that it actually worked, and you only pay for verified successes.

Get started freeSee live benchmarks

Drop-in for any OpenAI-compatible client

# any OpenAI-compatible clientnpm i @codai/sdkbase_url: https://ai.codai.ro/v1model: codai

Same request shape, one model name. No SDK rewrite.

Your toolsFrontier models✓ verified
codai

Works in

  • GitHub Copilot
  • Cursor
  • Claude Code
  • Codex CLI
  • OpenAI SDK
  • Vercel AI SDK

How it works

Three steps. No model management.

You keep the client you already use. codai does the routing, the verification and the accounting.

  1. 01

    Point your client at codai

    Set the base URL to ai.codai.ro/v1 and the model to codai. Copilot, Cursor, Claude Code, Codex CLI or any OpenAI SDK — nothing else changes.

  2. 02

    The router picks the frontier model

    Every request is classified and sent to the model that wins on that kind of task — Claude, GPT or Gemini — with automatic failover when a provider degrades.

  3. 03

    You confirm “did it work?” and pay per success

    A task is only billed when it is verified as successful — by tests, by the harness, or by your one-tap confirmation. Failed attempts cost you nothing.

What you get

A model router with an opinion about outcomes

Everything an agentic workflow needs, behind one name.

  • Frontier router

    Per-task model selection across Claude, GPT and Gemini, driven by open benchmarks and live production telemetry.

  • Tools up to L4

    HTTP, Python, shell and browser automation run server-side in sandboxes — so any client can use them, even without native tool support.

  • Persistent memory

    Facts, decisions and repo conventions survive across sessions and devices. Scoped per user, exportable, deletable.

  • MCP servers

    Attach your own Model Context Protocol servers once and every codai client gets the same tools.

  • Verified tasks, pay per success

    Outcome billing: a flat charge per verified successful task instead of a token meter you cannot predict.

  • Bring your own keys

    Use your existing Anthropic, OpenAI or Google keys through the same router and tooling, from €5 a month.

  • EU data, Cloud Run

    Hosted in europe-west1 on Google Cloud Run. Prompts and usage stay in the EU; no training on your data by default.

  • Open benchmarks

    Routing decisions are backed by public, reproducible benchmark runs. Check the numbers before you trust the router.

Model roster

Every frontier model, one endpoint

codai selects across Claude, GPT and Gemini. When a new model ships, we benchmark it and add it to the pool — you never rename anything.

Anthropic

3 models

  • Claude Fable 5

    Frontier model, autonomous coding

    large

  • Claude Sonnet 4.6

    Balanced quality and cost

    medium

  • Claude Haiku 4.5

    Ultra-fast, lightweight tasks

    small

OpenAI

3 models

  • GPT-5

    Frontier reasoning model

    large

  • GPT-5 mini

    Fast, cost-effective tasks

    medium

  • GPT-5 nano

    High-throughput micro-tasks

    small

Google

3 models

  • Gemini 2.5 Pro

    Long context, multimodal

    large

  • Gemini Flash 2.5

    Low latency, high throughput

    medium

  • Gemini Flash 2.0

    Sub-10 ms, experimental

    small

Use model="codai" in any OpenAI-compatible client. codai handles the rest.

Benchmarks

Don’t trust the router. Check it.

Every routing decision is grounded in public benchmark runs over real task suites — coding, agentic tool use, long context. Same harness for every model, results published as they land.

  • Open methodology and task sets
  • Paired comparisons, not single lucky runs
  • Refreshed as new models ship
Open the benchmarks

Pricing

Pay for tasks that worked

Every plan includes a monthly number of verified successful tasks. Beyond that, a flat €1.20 per verified success — never a token meter.

  • Free

    Try the router with real tasks.

    Free

    5 tasks included

    €1.20 per successful task beyond your plan

    Create account

  • Student

    For one developer shipping every week.

    €40/month

    33 tasks included

    €1.20 per successful task beyond your plan

    Choose Student

    Cancel anytime

  • Most popular

    Builder

    For builders running agents daily.

    €100/month

    83 tasks included

    €1.20 per successful task beyond your plan

    Choose Builder

    Cancel anytime

  • Pro

    For teams with production workloads.

    €300/month

    250 tasks included

    €1.20 per successful task beyond your plan

    Choose Pro

    Cancel anytime

Have your own provider keys? BYOK from €5/month with a 7-day trial.

See all plans

Trust

Built for European teams

Your prompts, keys and usage data are handled the way you would handle them yourself.

  • EU hosting

    Gateway, database and memory run on Google Cloud Run in europe-west1 (Belgium). Nothing is stored outside the EU.

  • No training on your data by default

    Prompts and completions are never used to train models unless you explicitly opt in from your account.

  • GDPR export and delete

    Export everything we hold about you, or delete your account and data, from your account page — no support ticket needed.

  • Passkeys and scoped keys

    Sign in with passkeys; mint API keys with per-key budgets and model scopes, and revoke them instantly.

codai

Start with five free verified tasks

No card needed. Point your client at codai and see where the router sends your first request.

Create your accountCompare plans