> ## Documentation Index > Fetch the complete documentation index at: https://fireworks.ai/docs/llms.txt > Use this file to discover all available pages before exploring further. # Account quotas > Configure Fireworks account-wide request limits, spending tiers, monthly budget caps, and on-demand GPU quotas to control usage and costs. Fireworks uses different controls for serverless and on-demand deployments. This page is the canonical reference for spending tiers, monthly spend limits, on-demand GPU quotas, and account-wide request limits. For serverless TPM and adaptive limits, see [Serverless rate limits](https://fireworks.ai/docs/serverless/rate-limits.md). ## Check your current limits View your account's current quotas and limits: ```bash theme={null} firectl quota list ``` This shows your rate limits, GPU quotas, spend limits, and usage across serverless and on-demand deployments. ## Spending tiers Your spend tier controls available quotas and capacity. For legacy self-serve postpaid accounts, it also determines the maximum monthly spend limit. Prepaid accounts can set a monthly spend limit independently of tier. | Tier | Criteria | Legacy Postpaid Max Monthly Spend Limit | | - | - | - | | Tier 1 | [Valid payment method and billing profile](https://fireworks.ai/billing) | \$50 | | Tier 2 | Spend or add \$50 in credits | \$500 | | Tier 3 | Spend or add \$500 in credits | \$5,000 | | Tier 4 | Spend or add \$5,000 in credits | \$50,000 | | Unlimited | [Contact us](https://fireworks.ai/company/contact-us) | Unlimited | > [!TIP] > > Add prepaid credits to unlock a higher tier. For example, adding \$100 moves you from Tier 1 to Tier 2. Your new tier activates within minutes. > [!NOTE] > > These spending tiers set the base [serverless TPM ceilings](https://fireworks.ai/docs/serverless/rate-limits.md) for the standard adaptive limit. [Elastic Burst](https://fireworks.ai/docs/serverless/rate-limits.md#elastic-burst) can temporarily put the current limit above that ceiling. The spend-limit maximum shown above applies only to legacy self-serve postpaid accounts. > > Fireworks operates on a pre-paid credits billing system. Contracted customers may have the option to move to post-paid billing — [contact our sales team](https://fireworks.ai/company/contact-us) to discuss your options. ### Training GPU quota Training jobs use training GPU quota (separate from on-demand deployment quota): * **No payment method:** 0 GPUs * **Payment method on file:** 32 GPUs of each type — H200, B200, and B300 Each accelerator type is granted independently (32 H200 plus 32 B200 plus 32 B300). Higher spending tiers do not increase this default. Before you start a job, use the [training cost estimator](https://fireworks.ai/docs/fine-tuning/cost-estimator.md) to size the run and confirm you have enough credits. If a job is rejected with HTTP 403 `quota_exceeded` or `tier_required`, add a payment method or free unused GPUs of that type, then resubmit. > [!NOTE] > > Need more than 32 GPUs of a type? [Reach out for enterprise support](https://fireworks.ai/contact-training) and we'll help size the right allocation for your workload. ### Enterprise accounts Enterprise accounts can configure monthly spend alerts, but those alerts do not pause service. They track the cost of Fireworks usage, including usage paid for with credits. Adding credits itself does not count as spend. For Enterprise quota details, see [Enterprise quotas](https://fireworks.ai/docs/faq/enterprise/service/quotas) or contact your account representative. ## Manage your quotas ### Account-wide request limits All API usage on your account shares a single request-throughput envelope: | Account state | Request-rate limit | | - | -: | | No payment method or no credits | 10 RPM | | Payment method and active credits | 6,000 RPM (maximum) | The **6,000 RPM** cap applies account-wide—it is **not** a separate serverless-only limit—and it is a **fixed** ceiling, not adaptive. Per-minute request volume above this cap is rejected (for example HTTP 429), regardless of your spending tier. ### Monthly spend limit Set a monthly spend limit that fits your needs and adjust it anytime. For prepaid accounts, the limit tracks the dollar value of your usage regardless of whether credits pay for that usage. Adding credits increases your available balance, but does not raise or reset your monthly spend limit. Auto Reload is separate from the spend limit. Auto Reload purchases credits when your balance is low; it does not change how much usage your account can accrue during the month. ### View and adjust your spend limit Check your current spend limit: ```bash theme={null} firectl quota list ``` Set a custom monthly spend limit: ```bash theme={null} firectl quota update monthly-spend-usd --value ``` For example, to set a \$200 monthly spend limit: ```bash theme={null} firectl quota update monthly-spend-usd --value 200 ``` ### Warnings and suspension By default, Fireworks sends a warning email when usage reaches 80% of your monthly spend limit. You can add other notification amounts from the Billing page. When usage reaches 100% of the limit, all API requests pause automatically across serverless inference, deployments, and training. Raise the monthly spend limit to resume usage. If your credit balance is also depleted, you must add credits as well. > [!NOTE] > > The suspension behavior described here does not apply to Enterprise accounts. Their monthly spend alerts are informational. ### On-demand deployment quotas On-demand deployments have GPU quotas instead of rate limits. GPU quota is scoped per placement (multi-region or single region). The defaults below apply to the **GLOBAL** multi-region only. For other placements, see [Regions](https://fireworks.ai/docs/deployments/regions.md). | GPU Type | Default Quota (GLOBAL) | | - | - | | Nvidia H100 | 16 GPUs | | Nvidia H200 | 16 GPUs | | Nvidia B200 | 16 GPUs | | Nvidia B300 | 16 GPUs | | Nvidia GB300 | 0 GPUs | | Nvidia A100 | 0 GPUs | | AMD MI325X / MI350X | 0 GPUs | | LoRAs (on-demand) | 100 | > [!TIP] > > Need more GPUs? [Contact us](https://fireworks.ai/company/contact-us) to request a quota increase. On-demand and dedicated deployments are **not limited by adaptive serverless TPM upper bounds**. If you receive HTTP **429** on those endpoints, it typically means **deployment saturation** (GPUs busy) rather than hitting a TPM tier cap. Requests still count toward [account-wide request limits](#account-wide-request-limits). See [understanding 429 errors](https://fireworks.ai/docs/guides/inference-error-codes.md#understanding-429-errors) for details and resolution steps. ### Account recovery If your account is suspended due to payment issues: 1. Go to [Billing](https://fireworks.ai/billing) 2. Resolve failed payment methods and add credits (or pay outstanding invoices for postpaid accounts) 3. Your account reactivates automatically within an hour > [!TIP] > > Still suspended after resolving payment issues? Contact support via [Discord](https://discord.gg/fireworks-ai) or email [inquiries@fireworks.ai](mailto:inquiries@fireworks.ai). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.