One APIEvery frontier modelPrivate & secured

Simon Says,The last AI integration you’ll ever write.

Simon picks the best model for you

api-call.ts

fetch("/v1/chat", {
method: "POST",
body: {
model:
input: "..."
})

fetch("https://api.geniuspro.io/v1/chat/completions", {
method: "POST",
headers: { "Authorization": "Bearer YOUR_KEY" },
body: JSON.stringify({
model:
messages: [{ role: "user", content: "..." }]
})
})

Every request is encrypted and private.

See How It Works

One OpenAI-compatible API connects you to every frontier model. When a new model ships, we wire it up — your code never changes. Simon routes each request to the best model available today, or pin any specific model directly. Private by default, never used for training.

One invoice. Unlimited API keys per workspace. SOC 2 & ISO 27001.

MicrosoftGoogleAmazonGroqNVIDIAIBMOracleCloudflareSupabaseVercelOpenAIAnthropic

US-based AI. Always encrypted.

One platform. Every top model.

Simon picks the right model for every request, automatically.

GeniusProSimon Says Models

OpenAIGPT Models

AnthropicClaude Models

GoogleGemini Models

MetaLlama Models

MicrosoftPhi Models

NVIDIANemotron Models

AWS

BedrockNova Models

Arcee AITrinity Models

PerplexitySonar Models

You

Hands-off by default.Hands-on when you need it.

Simon keeps your integration current for you. When you want to drive, every knob is yours.

We Maintain

Integrate once. Stay current.

Simon routes each request
Classifies the prompt and picks the best frontier model available today. Upgrades itself as better models ship.
New models live day one
Every new frontier model is wired up the day it ships, behind stable simon-says-* slugs. Your code keeps working; the model underneath gets better.
Automatic failover
If an upstream provider errors out or goes down, Simon reroutes to the next-best model with no operator intervention.

You Control

Full control when you need it.

Pin any specific model
Call gpt-5.4, claude-opus-4-7, gemini-3.1-pro, grok-3, or any direct slug when you need exact reproducibility.
Build your own CATs
Wrap a chosen model, system prompt, and tools into a stable cat:your-slug. You decide when the underlying model upgrades.
Cost caps & audit logs
Spending limits per workspace, revoke keys without redeploys, every request logged with its routing decision, cost, and latency.

Three ways to call a model.

Automatic routing, stable slugs, or direct model names — same API, same billing.

Automatic

Let Simon Decide

model: "simon-says"

Just send your prompt. Simon picks the best model for the task.

ChatCodingReasoningDesignAgentsVisionVoice

You don’t think about it. It just works.

Use GeniusPro Models

model: "simon-says-chat"

Pick a model by task. We keep it on the best version — your code never breaks.

simon-says-chatsimon-says-websearchsimon-says-codingsimon-says-coding-balancedsimon-says-coding-supersimon-says-coding-qualitysimon-says-reasoningsimon-says-designsimon-says-architecturesimon-says-architecture-prosimon-says-agentssimon-says-imagesimon-says-image-prosimon-says-visionsimon-says-voice-standardsimon-says-voice-qualitysimon-says-ttssimon-says-transcribe

Upgrades happen behind the scenes.

Direct Access

Pick Any Model

model: "claude-opus-4-7"

Call any model by name. One API, one bill, zero SDKs.

Claude Opus 4.7GPT-5.4Claude Opus 4.6Gemini 3 FlashLlama 4NemotronPhi-4Nova 2TrinitySonar

Full catalog at /v1/models.

What Will You Build?

From support bots to durable agents — one API handles the whole stack.

Simon + GeniusPro Models

AI-Powered Products

Ship AI features without juggling SDKs or worrying about lock-in.

Simon Auto-Routing

Customer Support Bots

Simon routes FAQs to cheap models and escalations to premium ones. Failover is automatic.

CATs — Custom AI Transformers

Custom AI with One Prompt

Write a prompt, pick a model, launch your own AI assistant. No code needed.

CAT Chains

Multi-Step AI Pipelines

Chain AI models, vision, and image gen into one workflow. Edit from the dashboard.

Tiger Mode

Production AI Agents

Long-running AI that retries on failure and picks up where it left off.

Full-Stack AI

Voice + Vision Apps

Voice, vision, image gen, and search — all through the same API.

Five Things No Other Platform Does.

Simon Orchestrator

One Model That Picks the Right Model.

Call "simon-says". A lightweight router classifies your prompt, picks the best specialist — coding, reasoning, design, agents — or runs several in parallel and merges the results. One price. No routing logic to maintain.

Auto-RoutingParallel DispatchCost OptimizationDirect Provider APIs

Self-Upgrading Models

One Name Forever. Always the Best Model.

simon-says-chat is a stable name, not a stable model. When a better frontier model ships, we swap it in. Your integration never changes. No deprecation notices. No version chasing.

simon-says-chatsimon-says-codingsimon-says-reasoningsimon-says-design

CATs & KITTNs

Multi-Step Workflows. One API Call.

A CAT (Custom AI Transformer) chains models, vision, image generation, and code transforms into a pipeline you invoke with cat:your-slug. Each step is a KITTN. Edit from the dashboard. No redeploy.

LLM StepsVision (SAM3)Image GenJS/Python Transforms

Tiger Mode

Durable Execution That Survives Failures.

Production AI can’t afford silent failures. Tiger mode persists progress after every step, retries on failure, and resumes where it left off. For high-stakes, long-running workflows where reliability is non-negotiable.

Step PersistenceControlled RetriesResume After FailureProgress Tracking

Full Stack AI

Models + Vision + Voice + Memory. One Endpoint.

Not just text. GeniusPro serves computer vision, real-time voice, persistent memory across sessions, integrated web search, and image generation — all through one unified API.

Computer VisionReal-Time VoiceSession MemoryWeb SearchImage Gen

Your Data Stays Private.

No training. PII redacted. Encrypted at every hop.

No Training On Your Data

Your prompts pass through to the model you chose and nothing is used to train anyone's models — not ours, not the upstream provider's.

PII Scrubbed From Our Logs

Emails, phone numbers, SSNs, credit cards, IP addresses and API keys are masked before requests land in our logs and stored history.

Opt-In Edge Scrubbing

Send X-GP-Scrub-Prompt: true (or scrub_pii: true in the body) and PII is masked before the request ever reaches the upstream provider.

US-Based AI Providers

Every model routes through a US-hosted company. Friendly to SOC 2, ISO 27001, and HIPAA-regulated workloads.

Always Encrypted

Requests are encrypted in transit over TLS and at rest in our managed infrastructure.

Fully Auditable

Every request logged. Usage, cost, and routing decisions tracked for your team.

Grows with You.

Solo developer, scaling startup, or multi-client agency — one platform, one bill, one audit log.

Organizations

One Roof, Full Control

Invite your team, manage billing, and see all AI usage from one dashboard.

Workspaces

Isolate Everything

Each workspace gets its own keys, balance, and logs. Bill per-project or per-client.

Client Portal

Give Clients Their Own View

$4,500/mo

Simon

$3,463/mo

You save

$1,037

Model	Direct cost	Simon Says	You save
Claude Opus 4.7	$10,000/mo	$3,463/mo	$6,53865%
Claude Opus 4.6	$10,000/mo	$3,463/mo	$6,53865%
GPT-5.4	$5,625/mo	$3,463/mo	$2,16238%
Claude Sonnet 4.6	$6,000/mo	$3,463/mo	$2,53742%
Gemini 3.1 Pro	$4,500/mo	$3,463/mo	$1,03723%

Assumes 75% input / 25% output. Simon Says blended rate across typical workload mix. Provider prices as of April 2026.

Simon Says,You’re Ready.

One API. Every frontier model. Private by default, always.