Build your own AI relay
One key for every AI. White-label, referral revenue, agent marketplace.
Model ecosystem examples · configure your own upstreams
One key connects the models you use
Claude / GPT / Gemini / Grok / DeepSeek — one console, every spend, every balance, every call detail. One invoice at month-end. Reconciliation time: gone.
- Real-time spend graph · drilldown by model and key
- Per-key quotas · auto throttle when exceeded
- Monthly invoice export · enterprise-ready
- Failed calls free · 100% cost traceability
Every major foundation model, one-click integrated. Full billing, invite, and reseller systems built in.
Ship your AI portal in 10 minutes
Your domain, your brand, your pricing. Referrals, multi-level resellers, agent payouts — all out of the box. You operate; AICodeDog runs the underlying stack.
- Custom logo · domain · colour system
- Per-model pricing · tier-based markup
- Invite links · referral commission · T+1 settlement
- Wallets · orders · invoicing fully managed
Operational scale, deep AI
One platform connects upstream routing, billing, branding, and reselling. These figures describe product structure, not historical traffic or an uptime guarantee.
Route Claude, OpenAI, Google, xAI, DeepSeek and other model traffic through one controlled gateway with clear provider configuration.
Centralize provider credentials, usage, and limits; the visible ledger makes comparisons easier.
See spend, latency, and request outcomes in the console, with alerts you can configure.
Upstream errors, timeouts, and filtered requests are zero-charge with automatic retry.
Keep per-call records and export billing data for reconciliation; enterprise invoicing depends on your deployment.
Pure usage-based pricing. No monthly minimum, stop whenever you want, no trailing fees.
Popular models
Curated top models in every category — start integrating in minutes.
Top-tier coding agent — the engine behind Claude Code; excels at long-context reasoning and tool use.
OpenAI’s flagship — strong agent capability, broad world knowledge, and reliable function calling.
Massive 2M-token context — excels at long documents, large codebases, and multimodal video scenarios.
King of price/perf — mainstream reasoning and coding ability at a fraction of the cost of comparable models.
Drop-in to every coding tool you use
/v1/chat/completions speaks OpenAI; /v1/messages speaks native Anthropic. Change one Base URL, paste a key, you are done.
From sign-up to first request in under 3 minutes
No phone calls, no waitlist, no procurement chasing. If you have a terminal, you can use AICodeDog.
- 01Sign up in 30 seconds
Email + magic link, no credit card needed. Every account gets trial credits — enough to ship a small side project end-to-end.
- 02Grab your AICodeDog key
Create as many per-application keys as you need from the dashboard, each with its own rate and spend limit.
- 03Change one env var
Point ANTHROPIC_BASE_URL (or OpenAI Base URL) at api.aicodedog.com — every model becomes callable.
Swap claude-sonnet-4-6 for any model shown in the model market — the rest of your code stays the same.
One platform, transparent billing, real service
Plug AI capability into your product without babysitting five upstream accounts at the same time.
OpenAI and Anthropic protocols passed through verbatim. Your existing SDK / CLI / Agent code learns nothing new.
Source delivery and managed setup use fixed prices; hosted API access stays usage-based with visible per-model rates.
Configure provider priority and model allow-lists; when several candidates are available, the relay can try the next one after a failure.
Tool calling, prompt caching, SSE streaming all work out of the box — no upstream feature is downgraded.
Per-application keys, rate limits, usage charts, per-call invoice export — all in one console.
Settled per call, failed calls free, hard cap on quota — you can never silently overdraw.
Inspect request latency and upstream health in the console; actual response time depends on provider and region.
SSE tokens passed through verbatim — no compromise on time-to-first-token.
Own brand domain, sub-accounts, custom pricing, Stripe Connect splits — all unlocked at the OEM tier.
Keep one usage ledger instead of reconciling several vendor invoices. Switch models freely and focus on shipping your product.
Generous Claude Code budgets for every engineer. One bill, one rotation, one audit trail.
Run your own brand on our infrastructure. Set your prices, collect your money, skip the ops.
Built around real workflows
The examples below describe supported workflows. Connect your own providers and validate them in a test environment before production.
“Swap one Base URL and API key, then keep using your existing OpenAI or Anthropic SDK.”
“Split keys, budgets, and usage by application so a team can reconcile spend from one console.”
“Configure brand, domain, and pricing for an OEM tenant; payment and upstream settings depend on your deployment.”
“Record provider health and try the next configured candidate when a request fails.”
“Review per-call billing records and validate the charging rule in a test environment first.”
“Use the curl, Python, and Node examples to validate one request before integrating the API.”
Frequently asked
Still have questions? Email [email protected] — every message gets a personal reply from an engineer.
AICodeDog primarily helps you buy or launch an AI relay stack. If you use the hosted API, one AICodeDog key can replace multiple vendor keys and billing dashboards while preserving OpenAI/Anthropic-compatible request flows.
Pure usage-based: every request is charged in real time against the upstream prompt/completion tokens and per-model unit price, with full per-call breakdowns in the console. Failed requests are not charged. Unit prices match upstream — markups are transparent.
Only the base URL and API key — your SDK, CLI, and agent code stay untouched. `/v1/chat/completions` is OpenAI-compatible (Cursor / Windsurf / OpenAI SDK work as-is). `/v1/messages` is the native Anthropic format (Claude Code / Anthropic SDK work as-is).
We alert at 80% and 100% thresholds. Once your balance hits zero, new requests return HTTP 402 with a top-up link — we will never silently overdraw your account. Auto-recharge limits are available to keep production traffic flowing.
Yes. Paste your own Anthropic / OpenAI key into the dashboard and AICodeDog becomes a pure routing layer — traffic hits your official accounts, we only handle aggregation, monitoring, and rate limiting. Ideal for teams with existing enterprise contracts.
By default we only log request metadata (tokens, model, latency) for billing and abuse prevention — full prompt and completion bodies are never persisted. You can opt into detailed audit logging from the dashboard if you need it.
Yes — unused balance can be refunded any time via your original payment method. Already-consumed real API calls follow the upstream provider policy (typically non-refundable).
Once you enable the OEM white-label tier we provide subdomains, sub-accounts, Stripe Connect revenue sharing, and a multi-tenant admin console — you can run your own AI API business on top of our infrastructure.
Run your own AI revenue
Sign up for free trial credits. Hook up Claude Code in 10 seconds, ship an OEM-branded portal in 10 minutes, and have referrals, resellers, agents, and invoicing all running within a month.