Both log model calls, track cost, version prompts and run a gateway. Helicone has been in maintenance mode since March 2026 and is not taking new cloud sign-ups, so for a new project Logfire is usually the practical choice of the two. If you already run Helicone, the useful question is what to move to and when.[3][4]
Helicone is an Apache-2.0 LLM request logger and AI gateway that has been in maintenance mode since March 2026, with new cloud sign-ups disabled; Logfire is Pydantic's actively developed OpenTelemetry platform with tracing, evals, prompt versioning and its own AI gateway. On 19 October 2026 Helicone said it is no longer part of Mintlify, which acquired it in March, and named no new owner. For a new deployment Logfire is usually the practical choice of the two, and existing Helicone users should plan a migration.
| Decision factor | Helicone | Logfire |
|---|---|---|
| Status | Maintenance mode since March 2026; cloud sign-ups disabled; said on 19 October 2026 it is no longer part of Mintlify | Pydantic, independent; SDK releases through October 2026 |
| Licence | Platform Apache-2.0; Rust gateway GPL-3.0, last updated November 2025 | SDKs MIT; platform closed source |
| Self-hosting | Free: Docker, Docker Compose or Helm | Enterprise only, on Kubernetes via Helm |
| Agent tracing | Sessions grouped by request headers; no span model for agent steps | OpenTelemetry spans for agent runs, tool calls and the surrounding application |
| Evals | Not an evaluation framework, per its docs; Python evaluators turned off in October 2026 | Offline experiments and live evaluations through pydantic-evals |
| Gateway | OpenAI-compatible, 100+ providers, failover, caching | Multiple providers, failover, load balancing, guardrails, spend caps |
| Plans | Hobby free, Pro $79 a month; new sign-ups disabled | Personal free; Team $49 a month; Growth $249 a month |
| Capability | Helicone | Logfire | Prefactor |
|---|---|---|---|
| Observe | |||
| LLM call tracing | ✓ | ✓ | ✓ |
| Multi-step agent tracing (tool calls, nested spans) | Partial | ✓ | ✓ |
| Cost tracking | ✓ | ✓ | ✓ |
| OpenTelemetry support | Partial | ✓ | ✓ |
| Evaluate | |||
| Offline evals (datasets, experiments) | Partial | ✓ | ✓ |
| Online evals on production traffic | Partial | ✓ | ✓ |
| LLM-as-a-judge scoring | Partial | ✓ | ✓ |
| Code-based evaluators | — | ✓ | ✓ |
| Human review and annotation | Partial | Partial | ✓ |
| Prompt versioning and playground | ✓ | ✓ | — |
| Act on findings | |||
| Runtime controls on live agent actions | Partial | Partial | ✓ |
| Gateway or proxy in the request path | ✓ | ✓ | — |
| Licence and deployment | |||
| Open source | ✓ | Partial | Partial |
| Self-hosting | ✓ | Partial | Partial |
Mintlify announced its acquisition of Helicone on 3 March 2026 and moved the product to a maintenance-only phase: patches, bug fixes and new model support, with help for customers migrating to other platforms. Cloud sign-ups have shown 'Sign ups are disabled' since the same day, and the last feature entry in the changelog dates from November 2025. On 19 October 2026 the site banner changed to say Helicone is no longer part of Mintlify, with details to come; no new owner or structure has been named.
The code is still moving, mostly defensively. In September 2026 Helicone fixed a platform-admin takeover and a cross-tenant read in its query language, reported through its disclosure programme. In October it shipped hardening for self-hosted installs, turned off Python evaluators and rate-limited LLM-as-a-judge calls. In May 2026 an AWS account lock took the dashboard and log processing down for about four days, while the proxy stayed up and queued logs were processed afterwards.
Logfire, by contrast, is in active development: the gateway moved into Logfire in March 2026, gained guardrails in July, and the Python SDK shipped version 5.2.0 on 13 October 2026. For a new deployment, that difference usually settles the choice before features do.
Helicone was built around its proxy. The AI gateway offers one OpenAI-compatible endpoint for more than 100 providers, with failover, caching, custom rate limits, bring-your-own keys and credits at 0% markup. Cost per request is exact through its model registry and best-effort for direct integrations, with pricing for over 300 models and cost alerts, but no hard spend caps. Its prompt-injection blocking works with OpenAI models only. Agent runs are grouped into sessions with request headers rather than traced as spans.
Logfire starts from OpenTelemetry tracing of the whole application, then adds a gateway. That gateway routes to OpenAI, Anthropic, Google, Bedrock, Groq and others, with bring-your-own keys, failover and weighted load balancing, hard spend caps, per-project, per-user and per-key limits, and per-rule guardrails that can flag, redact or block. It also runs evals through pydantic-evals, which Helicone's own docs say it does not do.
Map each Helicone feature you use before switching. Gateway traffic moves to Logfire's gateway once you confirm your providers are on its list and reissue keys from Logfire. Session headers give way to OpenTelemetry instrumentation through Logfire's integrations, which cover the OpenAI and Anthropic SDKs, LangChain, LangGraph, LlamaIndex, CrewAI, the Vercel AI SDK and LiteLLM, among others. Plan to re-create prompt versions and environment labels in Logfire's prompt versioning. Thumbs-up and thumbs-down feedback has a rough equivalent in Logfire's run annotations, which are still in beta. Cost views carry over in a different form: Logfire shows cost as a queryable trace column where telemetry includes token data, and its gateway meters spend per project, user and key.
Self-hosted Helicone users face a different trade-off. The Apache-2.0 code keeps running, but you now carry patching and the ownership question yourself. Logfire self-hosting is an Enterprise plan feature, so it is not a free like-for-like replacement for a self-hosted install.
As of 20 October 2026, Helicone lists Hobby free with 10K requests, 1 GB storage, one seat and 7-day retention; Pro at $79 a month with unlimited seats; and Team at $799 a month. With sign-ups disabled, new customers cannot start on any of them. Logfire Personal is free for 10M records a month with one seat, Team is $49 a month for five seats, Growth is $249 a month with unlimited seats, and extra records cost $2 per million. If you are moving off a paid Helicone plan, compare retention too: Helicone Pro keeps one month and Team three months, while Logfire Team keeps 30 days and Growth up to 90 days. See Helicone's pricing and Logfire's pricing.
We sell the layer this section describes. Read it with that in mind.
Stated plainly, since Prefactor publishes this page: Prefactor is an agent evaluation platform that scores each step of a run and acts on what it finds through runtime policies on tool calls and data access, approval holds for risky actions and a kill switch for a run, an agent or a team. Helicone's blocking covers prompt injection on OpenAI models, and Logfire's gateway covers model calls; neither covers tool actions or approvals. Prefactor has no prompt playground or prompt registry and its platform is hosted, so it does not replace a self-hosted Helicone install. See the full comparison of 15 platforms.
Reviewed against public product and documentation pages on October 20, 2026. If a vendor has changed a feature, product name, or positioning since then, send a correction and we will update it. Numbered source links in the page body point to the ordered sources below.
Prefactor scores each step of every agent run and can hold or stop an action when a score says it should not happen. Start on the free tier: 5,000 spans a month, no credit card required.
Prefactor helps teams observe, evaluate, and improve their AI agents in production, across every framework and provider.