- ✓ All 18 tools
- ✓ All slash commands
- ✓ 50 managed msgs/day
- ✓ Unlimited with your own key
- ✓ Per-project session memory
- ✓ Verification gates
- ✓ Everything in Free
- ✓ Unlimited ENSO + FLARE
- ✓ Generous PRIMUS + SHIORI quotas
- ✓ Auto-routing across all 4 models
- ✓ 7 parallel subagents
- ✓ Cross-session memory + hooks
- ✓ Plugins + MCP servers
- ✓ Priority support
- ✓ Shared team memory
- ✓ Team settings sync
- ✓ Usage dashboard
- ✓ Priority support
Same four models for everyone.
Free and Pro both get access to all four models. The difference is how many tokens you can spend per rolling window.
| Model | Role | Free / 5h | Pro / 5h |
|---|---|---|---|
| ▐ ENSO ▌ | Default | 200k tok | 2M tok |
| ▐ PRIMUS ▌ | Reasoning | 75k tok | 1M tok |
| ▐ SHIORI ▌ | Long context | 100k tok | 1.2M tok |
| ▐ FLARE ▌ | Fast | 500k tok | 5M tok |
Windows are 5-hour rolling buckets. Weekly caps are 5–8× the 5-hour cap.
Questions.
Do I need an account to start?
No. Install the CLI and start coding. The Free tier works with your own API key (Anthropic, OpenAI, Fireworks, OpenRouter, or any OpenAI-compatible proxy) — no signup needed. Sign in only if you want managed inference.
What counts as a "message" on the Free tier?
A user turn. One prompt → one round of tool calls + responses. Tool executions inside that round do not count. The 50/day cap resets at midnight UTC.
Can I switch between BYOK and managed?
Anytime. Run /provider in the CLI to switch. Keys stay encrypted on your machine. Managed inference requires arqzero login.
What happens when I hit a model quota?
You get a polite 429 and a hint to switch models. /model flare for fast-tier headroom, or wait for the 5-hour window to roll over.
Is my source code stored anywhere?
No. ArqZero is a thin client over the chat-completions API. Sessions save locally on your machine. Managed inference forwards prompts to the provider in real time; nothing is retained on our side beyond opaque token-count usage records.