Enterprise Billing
Enterprise Billing
The Enterprise plan primarily uses a per-seat billing model, tailored for enterprise-level teams, cross-department collaboration, high-security auditing, and complex multi-person R&D scenarios.
Enterprise Standard seats start at a minimum of 5 seats by default, and include a higher tier of permission controls, full-chain audit logging, and centralized billing management.
Enterprise Plans & Billing Rules
Enterprise plans are divided into the following three tiers based on organization size and technical depth:
| Plan Type | RMB Price (per seat/month) | USD Price (per seat/month) | Minimum Order | Core Benefits |
|---|---|---|---|---|
| Enterprise Standard | ¥699 / seat/month (¥6,990 billed annually) | $99 / seat/month (billed annually $990) | 5 seats minimum (from ¥3,495/month) | Includes advanced R&D collaboration, basic resource isolation, enterprise-grade permission firewall, and full-volume log auditing. |
| Enterprise Max | ¥489 / seat/month (¥4,890 billed annually) | $69 / seat/month (billed annually $690) | 6+ seats by application | For advanced technical teams that rely heavily on AI Swarms, offering a higher compute multiplier and dedicated support channel. |
| Enterprise Custom | Contact Sales | Contact Sales | 50+ seats | Supports hybrid-cloud/private deployment, a fully isolated credential center, custom team MCP, and dedicated model channel access. |
Note: Enterprise Max's per-seat price is intentionally lower than Enterprise Standard. Max includes everything in Standard and adds a higher compute multiplier and a dedicated support channel — the lower per-seat rate reflects its higher-volume commitment (6 seats minimum, by application), not a reduced feature set.
Enterprise Overage Billing Mechanism
Each Enterprise seat includes a high-spec base compute pack and service allowance. Once actual team usage exceeds the default range included in the plan, the excess is billed for actual consumption as follows:
- Model call consumption: Uniformly metered and audited via the enterprise-grade model gateway CodeGW, with usage beyond the base allowance billed on a tiered basis by actual token throughput.
- Agent resource consumption: Compute resources generated by Sub Agents or Swarms deployed within the enterprise while executing large-scale task graphs are dynamically deducted at a fixed points/hour rate — Swarm deployments are billed at 150 points/hour, and Sub (single Agent) deployments are billed at 10 points/hour.
BYOK (Bring Your Own Key)
The Enterprise plan fully supports BYOK mode.
Enterprise customers can seamlessly connect their own, controlled model service keys (for example, their own OpenAI, Anthropic Claude, Google Gemini, or enterprise-dedicated model API keys).
- BYOK cost attribution: When BYOK is enabled, token costs from underlying model calls are billed directly to the enterprise's own provider account, reducing platform pass-through costs.
- Platform management fees: Even when using the enterprise's own key, core platform-side functionality — Agent architecture orchestration, the resource provisioning wizard (Resource Grant), the team's high-risk operation approval firewall, full-chain auditing, and the management dashboard — continues to be billed or deducted as base platform resource usage according to the Enterprise plan's seat and platform rules.
Note: BYOK is not an Enterprise-exclusive capability — the personal plan's Free and Plus tiers also support connecting your own key for full feature access. What Enterprise adds on top of BYOK is team-level resource orchestration, the approval firewall, and full-chain auditing management capabilities — not BYOK itself.
Last updated on