Standard Compute
GlobalStandard Compute is a flat-rate multi-model LLM subscription for coding agents: each plan has a single monthly compute budget with no 5-hour windows, no weekly caps and no per-token fees — requests stop at the budget until renewal. Smart routing covers frontier and open-weight models (pick up to 5), stretching the budget up to ~70% further than paying providers directly. First-class integrations with OpenClaw, OpenCode, Kilo Code and Hermes Agent; EU/US inference regions. Individual plans currently start with 1.5x compute in the first month (limited-time launch offer); teams can create a company for free and mix plans per member.
Verified 2026-09-22
Standard Compute Plans
Starter (Monthly)
- ✓One agent at a time
- ✓1 API key
- ✓All models smart-routed; no 5-hour windows or weekly caps, requests stop at budget
- ✓7-day fair refund (minus compute used)
- ✓Works with OpenClaw, OpenCode, Kilo Code, Hermes Agent and any OpenAI-compatible client
Economy (Monthly)
- ✓Run agents in parallel
- ✓Shared execution pool
- ✓All models smart-routed; no 5-hour windows or weekly caps, requests stop at budget
- ✓7-day fair refund (minus compute used)
- ✓Works with OpenClaw, OpenCode, Kilo Code, Hermes Agent and any OpenAI-compatible client
Standard (Monthly)
- ✓Priority scheduling
- ✓Higher-capacity pool
- ✓All models smart-routed; no 5-hour windows or weekly caps, requests stop at budget
- ✓7-day fair refund (minus compute used)
- ✓Works with OpenClaw, OpenCode, Kilo Code, Hermes Agent and any OpenAI-compatible client
Pro (Monthly)
- ✓Highest priority scheduling
- ✓Maximum lane
- ✓All models smart-routed; no 5-hour windows or weekly caps, requests stop at budget
- ✓7-day fair refund (minus compute used)
- ✓Works with OpenClaw, OpenCode, Kilo Code, Hermes Agent and any OpenAI-compatible client
Pro Plus (Monthly)
- ✓Unlimited API keys
- ✓All models smart-routed; no 5-hour windows or weekly caps, requests stop at budget
- ✓7-day fair refund (minus compute used)
- ✓Works with OpenClaw, OpenCode, Kilo Code, Hermes Agent and any OpenAI-compatible client
Growth (Monthly)
- ✓High-capacity lane
- ✓More capacity for parallel agents
- ✓All models smart-routed; no 5-hour windows or weekly caps, requests stop at budget
- ✓7-day fair refund (minus compute used)
- ✓Works with OpenClaw, OpenCode, Kilo Code, Hermes Agent and any OpenAI-compatible client
Scale (Monthly)
- ✓Dedicated throughput
- ✓Priority support with a direct line to engineers
- ✓All models smart-routed; no 5-hour windows or weekly caps, requests stop at budget
- ✓7-day fair refund (minus compute used)
- ✓Works with OpenClaw, OpenCode, Kilo Code, Hermes Agent and any OpenAI-compatible client
Standard Compute FAQ
How much does Standard Compute cost?
There are 7 plans listed; the cheapest is Starter (Monthly) at $19/month.
What plans does Standard Compute offer?
Starter (Monthly) ($19/month), Economy (Monthly) ($39/month), Standard (Monthly) ($89/month), Pro (Monthly) ($249/month), Pro Plus (Monthly) ($499/month), Growth (Monthly) ($999/month), plus 1 more.
Which models does Standard Compute include?
The plans listed here cover GPT-6 Astra / Claude Fable 5.1 / Claude Opus 5 / Gemini 3.1 Pro Preview (frontier), GLM 5.3 / Qwen 3.8 27B / Kimi K3 / MiniMax M3 (open-weight), Gemini 3.8 Flash / Qwen 3.8 Flash / DeepSeek V4 Flash (efficient) and more. Check the provider's own docs for the authoritative list.
Does Standard Compute have a free tier?
None of the plans listed here is free; there is no ongoing free tier in this data.