Helicone Pricing in 2026 - Decoded, and Why the Meter Is a Mystery

Before decoding a single tier, the fact that changes the whole conversation. Helicone has been in maintenance mode since Mintlify acquired it on 3 March 2026. Only security patches, bug fixes and new-model support ship. No roadmap, no new features, and Mintlify is actively helping customers migrate off.

So this is a pricing guide with an asterisk. If you are already on Helicone and deciding when to leave, the numbers below help you plan. If you are pricing it fresh, the honest answer is do not - skip to the cheaper, transparent alternatives at the end. Either way, here is what the meter actually does.

The tiers, and the number nobody publishes

The plans are still live. Here is the shape.

Tier	Price	Included	Retention
Hobby	$0	10k requests/mo, 1 GB storage, 1 seat	7 days
Pro	$79/mo	10k free requests, then usage-based	1 month
Team	$799/mo	10k free requests, then usage-based	3 months
Enterprise	Custom	SAML SSO, on-prem, custom MSA	Custom

The quota unit is requests per month. The free Hobby tier is generous for prototyping - 10k requests, though only one seat and 7-day retention. Pro at $79/mo adds unlimited seats, alerts, reports, the HQL query interface and 1-month retention. Team at $799/mo is where SOC-2, HIPAA, five orgs and a dedicated Slack channel show up.

The problem is what happens above 10,000 requests. Both Pro and Team hand you 10k free requests and then say “usage-based.” Helicone does not publish the per-request overage rate. We could not find it from any primary or third-party source. That means the $79 Pro price is a floor, not a ceiling, and you cannot forecast your bill at any real volume from public information.

That was a weak spot even before the acquisition. Now that the product is frozen, there is nobody shipping a clearer pricing page to fix it. You are being asked to commit to a meter whose rate is undisclosed, on a tool whose owner is pointing you toward the exit.

Trying to estimate your bill (and why you can’t)

Normally this is where I work a real example. Say you run 500k requests a month on Pro. You would take the $79 base, subtract the 10k free requests, and multiply the remaining 490k by the overage rate.

Except there is no published overage rate to multiply by. The calculation stops at “490k requests times an unknown number.” That is the entire issue in one line. Any figure you have seen quoted is someone guessing.

The one part you can price with certainty is $0. The Hobby tier is free, and the self-hosted build is free too - Helicone is open source under Apache 2.0, so you can run it via Docker or Helm on Kubernetes with no license fee. But self-hosting a frozen codebase means you maintain a proxy, database, ingestion and scaling for a product that only gets security fixes. The subscription savings turn into on-call time for a tool with no future.

There is also the architectural cost baked into the proxy model. Helicone sits in the hot path of every LLM call - you change your base URL to route through it. If Helicone has downtime, your calls fail even when OpenAI or Anthropic are healthy, and every proxied call adds latency. Helicone cites around 10ms, which compounds for real-time apps. That is not a line item, but it is a cost.

Cheaper picks with pricing you can actually read

The good news is the two tools most teams move to are both cheaper than Pro and publish their overage rates in full.

Langfuse is the closest open-source replacement. Its Core cloud tier is $29/mo for 100k billable units, then a published $8 per 100k, stepping down to $6 at higher volume. The Hobby tier is free up to 50k units. And unlike Helicone, self-hosting Langfuse is a live, MIT-licensed product that keeps improving - only three features are enterprise-gated. At 1M events a month it runs about $101/mo managed, per Langfuse’s own pricing. The catch is operational: the v3 self-host needs Postgres plus ClickHouse, Redis and S3-compatible storage, so budget the ops time. But every number on that meter is public.

Opik is the cheapest managed cloud in the category. Its Pro tier is $19/mo for 100k spans, with a published $5 per additional 100k. The free cloud tier gives you 25k spans a month. Opik is Comet’s platform, Apache-2.0, and the self-hosted build ships the full feature set with unlimited spans, members and retention - no gates. The honest gotcha is per-seat pricing: the $19 headline is the small-team configuration, and seat costs climb as your team grows, so model that if you are big. But at least you can model it - the rate is on the page.

Both replace what Helicone did well - request logging and cost tracking - with pricing you can forecast and a codebase that is not frozen.

So what should you actually do?

• You are on Helicone today - use the runway to plan a migration. The tiers still work, but you are on a maintenance-mode product with an undisclosed overage rate. Do not scale further into it.
• You want the open-source replacement - Langfuse, free to self-host under MIT or $29/mo managed with a published $8 per 100k.
• You want the cheapest managed cloud with no ops - Opik at $19/mo for 100k spans, published overage, no license asterisk.
• You are pricing Helicone fresh in mid-2026 - don’t. A hidden meter on a frozen tool is the wrong place to start a production stack.

One honest note on the numbers. The Helicone tiers and both alternatives’ prices are from each vendor’s public pages, and this category ships breaking changes monthly - we re-verify every 30 days. The Helicone overage rate is flagged as not found because it genuinely is not published anywhere we could check. That, more than any headline price, is the reason to look elsewhere.