Helicone Pricing: Cost and Entitlement Worksheet

Helicone pricing has four commercial levels: Hobby, Pro, Team and Enterprise. The public page lists fixed monthly prices for Pro and Team, then applies usage pricing to paid self-service plans. A useful comparison therefore needs two ledgers. One captures Helicone subscription and usage. The other captures model-provider spend, operating work, retention, support and any separate request-policy layer.

TL;DR

• Hobby is free with 10,000 requests, 1 GB of storage, one seat and one organization.
• Pro is $79 per month and Team is $799 per month; both add usage-based charges after the included allowance.
• Retention, ingestion, API access, organization count, support and deployment rights change by plan.
• Helicone provides gateway routing and observability. Identity-bound authorization and independent policy evidence require a separate test.

Helicone pricing starts with entitlements

The Helicone pricing page lists Hobby at no charge, Pro at $79 per month, Team at $799 per month and Enterprise through sales. Hobby includes 10,000 requests, 1 GB of storage, one seat and one organization. Pro adds unlimited seats, alerts, reports and HQL. Team adds five organizations, a dedicated Slack channel and the compliance claims displayed on the plan card. Enterprise adds unlimited organizations, SAML SSO, a custom MSA, on-premises deployment, bulk cloud discounts and sales-led support.

Those prices are entry points. Complete workload totals also include usage because Pro and Team both carry a note that usage-based pricing applies. The page's comparison block shows 10,000 included requests on each paid tier, followed by usage charges, plus 1 GB of included storage. I would reject a quote comparison that copies only $79 or $799 into a procurement sheet. The cells to the right of that number determine the invoice and the operational fit.

Build a cost and entitlement worksheet

Use one row for each charge or right. The worksheet should show public evidence, contracted value, billing unit, forecast assumption, owner and acceptance test. Put these rows in the first tab:

• Base subscription: Hobby, Pro, Team or Enterprise, including billing cadence and contract term.
• Request usage: included requests, metered requests, forecast volume and the treatment of spikes.
• Storage: included capacity, chargeable capacity, retention period and deletion behavior.
• Organizations and seats: current need, growth assumption and separation between environments or business units.
• Ingestion and API access: published per-minute limits, required burst capacity and the response when a limit is reached.
• Support and deployment: support channel, service commitment, on-premises rights, implementation work and incident escalation.
• Provider charges: model tokens, embeddings and other provider costs kept outside the Helicone subtotal.

Put evidence beside each value. A clean worksheet should look slightly boring: one white row per cost owner, a source URL in the last column and no merged cells hiding two different meters.

Plan limits affect the operating model

The public comparison shows retention rising across the plans: seven days on Hobby, one month on Pro, three months on Team and forever on Enterprise. It also lists ingestion limits of 10 logs per minute on Hobby, 1,000 on Pro, 15,000 on Team and 30,000 on Enterprise. API access rises from 10 calls per minute on Hobby to 60 on Pro and 1,000 on Team. The page presents configurable retention and data export as Enterprise entitlements.

These limits need production tests. Replay a representative peak interval and confirm how the product behaves when log ingestion exceeds the plan limit. Export a known request and response, then verify the fields and timestamps. Delete a test record under the configured retention rule and document the result. The LLM observability guide explains the telemetry job; the worksheet makes its capacity and custody explicit.

Compliance labels also need contract detail. The page displays HIPAA and SOC 2 Type II in the comparison area, while SAML SSO, information-security reviews and customized MSAs appear at the Enterprise level. Ask which legal entity, hosted service, region, storage mode and features fall inside each assurance claim. A badge beside a tier cannot answer those scope questions.

Gateway value and provider cost stay separate

Helicone's AI Gateway overview describes one OpenAI-compatible API for more than 100 LLM providers, with routing, automatic fallbacks and unified observability. The page says Helicone credits carry zero markup and also supports bringing provider keys. It records metrics, costs and errors as the gateway returns a provider response.

Keep three cost owners separate in the workbook. Helicone owns the subscription and its usage meters. The model provider owns token and service charges. Your platform team owns deployment, integration, incident response and any customer-operated infrastructure. If caching changes provider spend, record cache hit assumptions and verify them with a fixed replay rather than treating a calculator estimate as a guaranteed saving.

Helicone is a sensible fit when routing and observability are the primary procurement job. The AI gateway guide gives the wider category context. My preference is to buy that job on its own merits and refuse to let a gateway line item absorb unrelated security claims.

Compare the authenticated request-policy boundary

Gateway placement creates useful request telemetry, but telemetry alone does not prove authorization. For one deliberately routed HTTP model call, capture the originating identity, role or delegated authority, prompt classification, policy version, destination, outcome, timestamp and integrity evidence. Change one meaningful attribute and repeat the call. The expected decision should change for a reason the reviewer can retrieve.

DeepInspect operates on that authenticated user or agent-to-LLM HTTP boundary. Helicone's cited pages focus on provider access, routing, fallbacks, cost and observability. Direct SaaS sessions, browser activity, local execution and tool permissions sit outside DeepInspect unless the relevant model request crosses its proxy. The AI policy enforcement guide covers this narrower control surface.

A coexistence design can route a model call through request policy and then through a gateway, provided the team defines timeout ownership, retry behavior, identity propagation and log correlation. Record both decision IDs in the worksheet. Force a provider timeout, a policy denial and a logging failure separately. That sequence tells you which component made each decision and which record survived.

DeepInspect

DeepInspect evaluates authenticated HTTP model calls deliberately routed through its stateless proxy. It applies policy to caller identity, route, model destination and prompt data before forwarding allowed traffic, then writes a per-decision record that can be correlated with Helicone's routing and observability record.

Helicone subscription, usage, provider routing, caching and telemetry remain separate cost and operating lines. Direct browser sessions, native SaaS AI, local execution and tool authorization remain outside DeepInspect unless the model request crosses the inspected route. Book a technical deep dive at deepinspect.ai.