Reeve.LLM

What it does

• To use the LLM feature, an app first needs to have it explicitly enabled by its owner, and then must present a valid access key when making requests—having just the key alone isn't enough. Once both are in place, the app can securely request AI-generated completions through our API. A separate usage-reporting endpoint is more permissive, but we independently verify any cost information submitted by unauthenticated sources.
• Reeve provides a single, consistent way to generate AI responses, whether you need the full answer at once or want it streamed back as it's generated. It supports tool use within conversations, giving your application the same reliable capabilities used across all of Reeve's own products. This means any application can access the same high-quality AI experience, regardless of how it's set up.
• Reeve.LLM usage is billed against your organization's shared credit pool, so everyone on your team draws from the same balance. If that pool runs out, any new requests will be declined until more credits are added.
• You can optionally set spending limits on your account—covering your whole organization or narrowed down to a specific product, user, or API key—on a daily or monthly basis. If a limit is reached, further requests will be paused with a clear notification until the next period begins. By default, no limits are applied unless you choose to set them.
• You can check your current credit balance at any time, along with how much you've spent and how much remains in each of your spending limit periods. This gives you clear visibility into your usage and remaining budget before you run out of credits.
• Reeve.LLM can remember and reuse the shared setup from your earlier requests for a short time (a few minutes, or longer if you specify). This means repeated or follow-up requests run faster and cost less, since the system doesn't have to reprocess the same information each time.

Getting started

1. Have the app owner enable the llm capability for your host app. It is not on by default, and a host key alone is not enough without it.
2. Authenticate with your Reeve host key (X-Reeve-Host-Key) to call the Reeve.LLM surface.
3. Make your first request and check the response, then wire it into your app.

Limits & notes

• Usage is metered against your Reeve credit balance via llm_cost_cents; see the pricing page for current rates.
• Requests authorize on your organization; a valid host key scoped to this capability is required.

FAQ

What does Reeve.LLM do?

To use the LLM feature, an app first needs to have it explicitly enabled by its owner, and then must present a valid access key when making requests—having just the key alone isn't enough. Once both are in place, the app can securely request AI-generated completions through our API. A separate usage-reporting endpoint is more permissive, but we independently verify any cost information submitted by unauthenticated sources.

What else should I know about Reeve.LLM?

Reeve provides a single, consistent way to generate AI responses, whether you need the full answer at once or want it streamed back as it's generated. It supports tool use within conversations, giving your application the same reliable capabilities used across all of Reeve's own products. This means any application can access the same high-quality AI experience, regardless of how it's set up.

How is Reeve.LLM billed?

Usage is metered against your Reeve credit balance; see the pricing page for rates.

Sources

8 of 15 claims on this page are grounded in the Reeve.Knowledge product corpus. The rest are product/catalog reference copy with no Knowledge card behind them, listed as uncited below.

• does.0 — [source]
• does.1 — [source]
• does.2 — [source]
• does.3 — [source]
• does.4 — [source]
• does.5 — [source]
• faqs.0.a — [source]
• faqs.1.a — [source]
• overview — uncited (reference copy)
• how_to.0 — uncited (reference copy)
• how_to.1 — uncited (reference copy)
• how_to.2 — uncited (reference copy)
• limits.0 — uncited (reference copy)
• limits.1 — uncited (reference copy)
• faqs.2.a — uncited (reference copy)