Model gateway for agents

Tekiz model gateway

Hosted model gateways are useful when you want one marketplace-style API. Tekiz is built for teams and agents that want one endpoint too, but with local-first control, BYOK routing, private provider credentials, Ollama support, health-aware fallback, and an optional hosted account layer.

Create hosted API key Migration guide Fusion compatibility Compare plans Compare OpenRouter Run locally View public edge status Estimate routing savings Max implementation review
Recover API key

Create an account now. Connect an authorized provider or add billing when you are ready to route.

Compliance boundary: Tekiz routes access customers are authorized to use. The hosted service is account, billing, routing, telemetry, and reliability infrastructure, not a promise to bypass provider terms.

Evaluating OpenRouter specifically? Read the focused Tekiz vs OpenRouter comparison for marketplace API vs local-first routing tradeoffs.

Pick Tekiz when keys should stay under your control.

Default deployments keep provider credentials on your machine, server, Tailnet, or private router. Hosted account access can manage keys and billing without becoming the only custody model.

Pick Tekiz when agents need fallback policy.

Route by task, model profile, provider health, latency, local-vs-cloud preference, and failure history. Use one endpoint across Codex, OpenClaw, Cursor, Aider, Continue, and OpenAI-compatible tools.

Pick Tekiz when local models matter.

Local Ollama and Ollama Cloud routes can sit beside OpenAI, Anthropic, Gemini, NVIDIA NIM, endpoints, and other authorized providers.

Comparison

Need Tekiz Hosted model gateway
Credential custody Local-first by default. Provider keys can remain on customer-controlled routers. Typically centralized through the hosted API account.
Agent harness support OpenAI-compatible, Anthropic-compatible, Codex, OpenClaw, Cursor, Aider, Continue, OpenHands, and harness fallback docs. Strong hosted OpenAI-compatible API story.
Local models First-class. Route local Ollama and local/cloud hybrid profiles beside hosted providers. Usually focused on remote hosted model providers.
Provider selection Policy, health, latency, capability, task intent, route profiles, and fallback chains. Model catalog selection and provider routing through hosted infrastructure.
Fusion / compound answers Premium Pro/Max route. Use sage-router/fusion or the sage-router:fusion server tool to derive explicit bounded work locally, run independent tasks in parallel, and verify the result. Compound routing can expose a hosted multi-model workflow.
Resilience model Tailnet and edge routing can select the healthiest Sage Router installation and keep private routers behind the public API edge. Depends on the hosted provider platform and upstream provider health.
Hosted SaaS Hosted signup, Stripe checkout, API keys, usage quotas, status page, analytics, and Supabase-backed auth. Mature hosted account and billing experience.
Best fit AI agents, teams with existing provider access, local-first deployments, private routing, and policy control. Hosted model marketplace access with minimal local infrastructure.

How to switch

Use one OpenAI-compatible base URL.

For hosted access, create an account and key, then point clients at the public Tekiz edge. For local-first access, run the router on port 8790 and keep provider credentials on your machine or Tailnet host. 

export OPENAI_BASE_URL="https://api.tekiz.ai/v1"
export OPENAI_API_KEY="sk_sage_..."

curl "$OPENAI_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sage-router/frontier",
    "messages": [{"role": "user", "content": "Route this well"}]
  }'
Copy setup first. Paste the bundle into your current OpenAI-compatible client, then replace the placeholder with your generated sk_sage_* key.

Monetization fit

Hosted convenience without losing local control.

Tekiz can sell hosted API keys, usage analytics, fallback reliability, team policy sync, and support while keeping the core open-source router useful for local and private deployments.

  • Lite: simple hosted routing and usage visibility for individual agent workflows.
  • Pro: frontier routing profiles, Fusion when capacity is live, better fallback policy, and analytics for serious development work.
  • Max: high-volume automation, priority fallback, Fusion when capacity is live, and team use.
  • Max implementation review: request migration help, private deployment support, or future one-subscription model access only after provider authorization, margin, quota, operator-audit, and abuse-control gates are ready. Request private-beta review.