Standard Compute is an LLM gateway for AI coding agents: one OpenAI- and Anthropic-compatible API that routes each request to a cost-efficient model, paid with a monthly plan that has a fixed compute budget instead of per-token billing.
https://api.stdcmpt.com/v1https://api.stdcmpt.com (no /v1)https://us.stdcmpt.com/v1standardcompute| Plan | For | Price / month | Compute budget |
|---|---|---|---|
| Starter | Individual | $19 | $20 |
| Economy | Individual | $39 | $41 |
| Standard | Individual | $89 | $95 |
| Pro | Individual | $249 | $269 |
| Pro Plus | Business | $499 | $549 |
| Growth | Business | $999 | $1,119 |
| Scale | Business | $2,499 | $2,849 |
| Business | Business | Custom | Custom |
Prices in USD. Budgets are spent at each model's per-token rate. Trial: $0.25 of compute for eligible new accounts, no card. Current offers on pricing.
| Agent | How it connects |
|---|---|
| Claude Code | Anthropic Messages API (/v1/messages). Set ANTHROPIC_BASE_URL to https://api.stdcmpt.com, or run standardcompute launch claude. |
| Codex CLI | OpenAI Responses API (/v1/responses). Run standardcompute launch codex. |
| OpenCode | Listed natively through models.dev. Run /connect and choose Standard Compute. |
| OpenClaw | Run standardcompute launch openclaw. |
| Hermes Agent | Run standardcompute launch hermes. |
| Cline | "OpenAI Compatible" provider, base URL https://api.stdcmpt.com/v1, model standardcompute. |
| Kilo Code | "OpenAI Compatible" provider, base URL https://api.stdcmpt.com/v1, model standardcompute. |
| Roo Code | "OpenAI Compatible" provider, base URL https://api.stdcmpt.com/v1, model standardcompute. |
| Pi | OpenAI-format provider in ~/.pi/agent/models.json with the base URL. |
| Oh My Pi | openai-completions provider in ~/.omp/agent/models.yml with the base URL. |
Endpoints: /v1/chat/completions, /v1/responses, /v1/messages, /v1/completions, /v1/models. More setup guides on integrations.
From the gateway's model catalog on 13 October 2026. Live list on models.
No. Every plan has a fixed monthly compute budget, shown live in the dashboard. Requests run at full speed until the budget is used, then return HTTP 402 with the code budget_exhausted until the billing period renews. To keep going sooner you can upgrade, or on Economy and above turn on Pay as you go. There are no 5-hour or weekly windows inside the month.
No. Standard Compute is a separate API service with its own API key and its own monthly budget. It does not use or resell your Claude Pro, Claude Max or ChatGPT login. Claude, GPT and other models are reached through upstream model providers and billed against your Standard Compute plan.
No. You do not rent GPUs, servers or clusters. Standard Compute is a model API: you send requests and get model responses. It is also not related to Databricks' "standard compute" cluster tier.
Claude Code through the Anthropic Messages API (ANTHROPIC_BASE_URL=https://api.stdcmpt.com), Codex CLI through the OpenAI Responses API, OpenCode through its built-in models.dev provider list, OpenClaw and Hermes Agent through the standardcompute launcher, and Cline, Kilo Code, Roo Code, Pi and other OpenAI-compatible tools with the base URL https://api.stdcmpt.com/v1.
Yes. Send the model standardcompute and the router picks among the models you enabled (up to 5), sending simple work to efficient models and hard work to stronger ones. Or send an exact model ID to pin that model for the request.
The account database runs in the EU, on AWS eu-central-1 (Frankfurt) via Supabase. Prompts are sent to upstream model providers to generate a response; by default they can run anywhere, and an opt-in Europe or United States Compliance rule limits them to provider endpoints in those regions. The Terms are governed by Norwegian law, with the courts of Oslo as venue.
Yes. A company or agency workspace is free to create. Each API key carries its own plan, approved and paid by an admin, or runs on Pay as you go from a prepaid balance at usage plus 15%, with spend limits per key. Admins can set Europe or United States rules for the whole company, a team or a single key.