Per-token pricing you can read. Plans that scale with your team.
Every model has a clear input and output price per million tokens. No hidden markups, no per-request surcharges, no minimum commitments. Pick the plan that matches your workspace needs — model pricing stays the same across all tiers.
Starter — $29/mo
$29/month
- 1 workspace, 3 API keys
- All supported models (DeepSeek, Qwen, Kimi, GLM, MiniMax)
- Community support
- Usage dashboard and spend alerts
- Cost pass-through model pricing — pay only for tokens used
For prototypes and early production workloads that need a clean OpenAI-compatible entry point with transparent billing.
Team — $99/mo
$99/month
- Everything in Starter, plus:
- Up to 5 workspaces, 20 API keys
- Role-based access control (RBAC)
- Approved model routes per workspace
- Shared budgets and spend limits
- Priority email support
For organizations with multiple workspaces, role-based access, approved routes, and shared budgets that need governance without enterprise overhead.
Enterprise — Custom
Custom / contact sales
- Everything in Team, plus:
- Unlimited workspaces and API keys
- Regional policy enforcement (US, EU)
- Security review and SOC 2 alignment
- Residency requirements and data controls
- Negotiated commercial terms and volume pricing
- Dedicated support and SLA
For regional policy, security review, residency requirements, and negotiated commercial controls that need enterprise-grade governance.
Which plan fits your workload?
Model pricing is the same across all plans — you choose your tier based on the workspace features and governance controls you need. Here is a practical guide:
| Starter ($29/mo) | Team ($99/mo) | Enterprise (Custom) | |
|---|---|---|---|
| Workspaces | 1 | Up to 5 | Unlimited |
| API keys | 3 | 20 | Unlimited |
| RBAC & approved routes | — | Included | Included |
| Regional policy (US/EU) | — | — | Included |
| Best for | Solo devs, prototypes, early production | Teams with multiple projects and budget controls | Orgs needing compliance, residency, SLA |
Model pricing at a glance
Pay only for the tokens you use. All prices per million tokens. Model pricing is the same across Starter, Team, and Enterprise plans — you choose your plan based on workspace features, not model access.
| Model | Input /M tokens | Output /M tokens | Details |
|---|---|---|---|
| DeepSeek Chat | $0.30 | $1.19 | View details → |
| Qwen Plus | $0.30 | $1.19 | View details → |
| MiniMax M3 | $0.31 | $1.25 | View details → |
| GLM-5 | $0.60 | $2.68 | View details → |
| DeepSeek Reasoner | $1.79 | $3.57 | View details → |
| Qwen3.8 Max | $1.79 | $5.36 | View details → |
| Kimi K3 | $2.98 | $14.88 | View details → |
All model prices are cost pass-through. See individual model pages for full pricing details, context windows, and API examples. For a comprehensive model comparison, read our Best Chinese AI Models 2026 guide.
How billing works
BestLLM uses a reservation-based billing system. When your request arrives, the gateway reserves the estimated cost based on input tokens and the model price. As the response streams, the reservation is captured for the actual token count. When the response completes, the reservation is released and the final charge is recorded in the ledger.
This means you are billed for exactly the tokens that flow through the gateway — not estimates, not rounded-up aggregates. Every ledger entry is append-only and includes a versioned price snapshot from the moment of dispatch, so your invoices are always reconcilable to individual requests.
Provider cost and customer charges remain separate by design. The price you see is the price the gateway bills. There is no hidden markup layer between what providers charge and what you pay.