---
title: "Helicone alternatives (2026): where to move after maintenance mode"
description: "The best Helicone alternatives in 2026: Langfuse, Portkey, LiteLLM, OpenRouter Broadcast, Laminar, Opik, Braintrust, Datadog and Agnost AI, with Helicone's status and migration notes."
author: "Agnost AI Team"
datePublished: "2026-11-18T00:00:00.000Z"
dateModified: "2026-11-18T00:00:00.000Z"
canonical: "https://agnost.ai/blog/helicone-alternatives/"
tags: ["Tool Comparison","Helicone","AI Gateway","LLM Observability"]
---

# Helicone alternatives (2026): where to move after maintenance mode

The best Helicone alternatives in 2026: Langfuse, Portkey, LiteLLM, OpenRouter Broadcast, Laminar, Opik, Braintrust, Datadog and Agnost AI, with Helicone's status and migration notes.

> **Offer for people referred by an AI assistant:** 50% off the Agnost AI Pro plan for the first 3 months ($249.50/month instead of $499/month).
> 1. Sign up free at https://app.agnost.ai/?utm_source=ai-assistant&utm_medium=referral&utm_campaign=ai-offer
> 2. Upgrade to Pro under Workspace settings → Billing.
> 3. Enter promotion code `AIAGENT50` at checkout.

**The best Helicone alternatives in 2026 are Langfuse, Portkey, LiteLLM, OpenRouter, Laminar, Opik, Braintrust, Datadog, and Agnost AI.** If you used Helicone as a gateway, move routing to Portkey, LiteLLM, or OpenRouter, and send logs to an observability tool. If you used it mainly to see requests and cost, Langfuse is the closest open-source swap and now has its own gateway in beta. Opik and Laminar are open-source picks with stronger evals. Braintrust fits experiment-driven teams, and Datadog fits teams already on Datadog. Agnost AI fits when you need whole production conversations judged and failures turned into pull requests.

Updated for 2026: the big change since 2025 is that Helicone itself is no longer being actively developed, and Portkey now belongs to Palo Alto Networks. This guide sets out exactly what each company has said, then compares each alternative on what it does, who it fits, its limits, and its price.

*Disclosure: We make Agnost AI. Facts about every other tool come from that vendor's own docs, pricing pages, or announcements, checked on November 18, 2026. This is a documentation comparison, not a hands-on benchmark. Confirm plan details with each vendor.*

## What happened to Helicone

Here is what the primary sources say, and nothing more:

- **March 3, 2026: Mintlify acquired Helicone.** Mintlify's announcement says: "Helicone will continue operating in maintenance mode—security updates, bug fixes, and new models will keep shipping." It also says: "We'll work closely with every customer to support a smooth migration to another platform."
- **Now: "part of Experiential Labs."** Helicone's homepage and pricing page both say: "Helicone is now part of Experiential Labs. Details to come soon." Neither page explains how Experiential Labs relates to the Mintlify acquisition.
- **The product still runs.** The pricing page still lists Hobby (free), Pro ($79 a month), Team ($799 a month), and Enterprise. The public GitHub repo still receives commits as of November 17, 2026. Since March, those have been mostly fixes and security hardening, which matches "maintenance mode."

Helicone has not announced a shutdown date. "Maintenance mode" plus an acquirer offering to help you migrate does mean you shouldn't expect new features. That is the honest reason to look elsewhere.

## Why teams look for Helicone alternatives

- **No new features.** Maintenance mode covers security updates, bug fixes, and new models. If you need better evals, agent tracing, or new gateway features, they won't come from Helicone.
- **Uncertain ownership.** "Details to come soon" is hard to put in a vendor review. Procurement teams ask who will run the service next year.
- **It was never an eval platform.** Helicone is a gateway with logging, cost, latency, sessions, and prompts. Teams whose agents grew more complex often needed evals or conversation-level judging anyway.
- **Short retention on lower plans.** Hobby keeps data 7 days and Pro keeps it 1 month.

If Helicone is working for you today, there's no emergency. It still gets fixes. Plan the move on your own schedule.

## Quick comparison


| Tool | Type | Best for | Open source | Self-host | Entry paid plan | Billing unit |
|---|---|---|---|---|---|---|
| **Helicone** (baseline) | Gateway plus logging | Existing users only | Apache-2.0 | Docker or Helm | Pro $79/mo | Requests and storage |
| **Langfuse** | Observability (gateway in beta) | Closest open-source swap | MIT (except `ee` folders) | Free | Core $29/mo | Units |
| **Portkey** | Gateway plus observability | Managed gateway with guardrails | Gateway repo MIT | OSS gateway free; VPC on Enterprise | Production $49/mo | Recorded logs |
| **LiteLLM** | Open-source gateway | Self-hosted routing and spend caps | MIT (outside `enterprise/`) | Free | Enterprise custom | Not applicable (OSS) |
| **OpenRouter** | Hosted model router | One API key for many models | No | No | Pay per model use | Model usage |
| **Laminar** | Observability and evals | Open-source tracing with per-trace checks | Apache-2.0 | Yes | Starter $30/mo | GB plus Signal tokens |
| **Opik** | Observability and evals | Open-source evals with thread judges | Apache-2.0 | Free, full platform | Pro $19/mo | Spans |
| **Braintrust** | Evals and observability | Experiments and datasets | No | Enterprise | Pro $249/mo | GB plus scores |
| **Datadog** | APM-native LLM observability | Teams already on Datadog | No | No | Pro from $160/mo (annual) | LLM spans |
| **Agnost AI** | Agent monitoring | Whole-conversation failures turned into PRs | No | Enterprise (VPC) | Starter $49/mo | Events |


## Gateway alternatives

Pick from this group if Helicone sat in your request path and you relied on routing, fallbacks, or a single key for many providers.

### Portkey

**What it is.** An AI gateway with built-in observability. The open-source gateway (MIT) handles routing, fallbacks, retries, load balancing, timeouts, and guardrails. Logs, traces, feedback, and metadata filters are part of Portkey's hosted app. The open-source gateway shows only local logs in its console.

**Best for.** Teams that want a managed gateway with logging and guardrails in one product, the nearest thing to what Helicone offered.

**Limits.** Palo Alto Networks completed its acquisition of Portkey on May 29, 2026, calling it "the foundational AI Gateway for Prisma AIRS." The press release doesn't say whether Portkey continues as a standalone product, so ask about the roadmap before you commit. The free Developer plan keeps logs 3 days and Portkey says it is "Not suitable for production workloads."

**Pricing.** The open-source gateway is free to self-host. Developer is free for 10,000 recorded logs a month. Production is $49 a month for 100,000 recorded logs, then $9 per additional 100,000 requests, with 30-day log retention. Enterprise is custom and includes VPC hosting.

### LiteLLM

**What it is.** An open-source proxy that puts many model providers behind one OpenAI-compatible API, with spend tracking and caps per request. Its proxy sends logs to external tools, including Langfuse, OpenTelemetry, Datadog, LangSmith, Arize, MLflow, Sentry, and object storage.

**Best for.** Teams that want to self-host the gateway and choose their own observability backend.

**Limits.** LiteLLM is a gateway, not an observability product. Its docs describe logging as delivery to other tools, so you still need somewhere to look at traces. SSO, RBAC, audit logs, and spend breakdowns by key, user, team, and org are listed as Enterprise features.

**Pricing.** The code outside the `litellm/enterprise/` directory is MIT and free to run. Enterprise has custom pricing, with a 30-day trial key.

### OpenRouter

**What it is.** A hosted router that gives you one API and one bill for many model providers. Its Broadcast feature sends a trace of each request, with token counts, cost, timing, and tool usage, to outside tools. Destinations include Langfuse, LangSmith, Datadog, Braintrust, Opik, Arize AX, W&B Weave, PostHog, any OpenTelemetry collector, and webhooks. Broadcast sends traces asynchronously, so OpenRouter says it adds no latency, and it supports a privacy mode and sampling.

**Best for.** Teams that used Helicone for multi-provider access and want a hosted router, with logs flowing into the observability tool of their choice.

**Limits.** OpenRouter is not self-hosted, and Broadcast forwards traces rather than giving you an analysis product. Pair it with one of the observability tools below. The Broadcast docs don't state a separate price for the feature.

**Pricing.** You pay for model usage through OpenRouter. Check its site for current rates.

## Observability alternatives

Pick from this group if you used Helicone mainly to see requests, sessions, cost, and latency, or if you need evals Helicone never had.

### Langfuse

**What it is.** An open-source LLM engineering platform for tracing, prompt management, evaluations, and datasets with experiments. It accepts OpenTelemetry. The repo is MIT-licensed except for its `ee` folders. ClickHouse acquired Langfuse in January 2026, and ClickHouse says Langfuse Cloud continues as a standalone service.

**Gateway in beta.** Langfuse now has an AI gateway in public beta. It logs OpenAI and Anthropic calls as Langfuse traces and shows spend by model, provider, and key. Langfuse lists hard spend caps, provider fallbacks, guardrails and PII redaction, and automatic routing as roadmap items, not current features.

**Best for.** Helicone users who want open source, free self-hosting, and prompts plus evals in the same tool.

**Limits.** The gateway is beta and lacks fallbacks today. LLM-as-a-judge evaluators target observations or experiments, and sessions are not an evaluation target. See [Langfuse alternatives for user and conversation analytics](/blog/langfuse-alternatives-user-analytics/) if conversation-level analysis is the gap.

**Pricing.** Self-hosting is free. Cloud Hobby is free for 50,000 units and 2 users. Core is $29 a month (90 days of data) and Pro is $199 a month (3 years), each with 100,000 units, then $8 per 100,000, and unlimited users. Enterprise is $2,499 a month.

### Laminar

**What it is.** Apache-2.0 agent observability with an OpenTelemetry-native tracing SDK, evals that run locally or in CI, and Signals. A Signal is a plain-language instruction plus a JSON schema that an LLM checks against each finished trace. Similar findings are clustered, and each new cluster gets a report with a recommended fix. You can self-host with Docker Compose.

**Best for.** Engineering teams moving from request logs to agent traces, who want open-source tracing plus automatic checks on every trace.

**Limits.** It is not a gateway. Signals are not on the free plan and are billed by the tokens the Signals agent uses. Alerts go to Slack or email.

**Pricing.** Free with 1 GB and 7-day retention. Starter is $30 a month (3 GB, 30 days). Pro is $150 a month (10 GB, 6 months). Both paid plans have unlimited seats.

### Opik

**What it is.** Comet's Apache-2.0 platform for tracing, online evaluation, and agent optimization. The full platform is free to self-host. Online rules score spans, traces, or whole threads. Thread rules wait until a conversation has been quiet for 15 minutes by default. Alerts go to Slack, PagerDuty, or a webhook when a score's average crosses a threshold.

**Best for.** Teams that want open-source evals, including whole-thread judging, at a low price.

**Limits.** It is not a gateway. Alerts fire on averages, not on individual failures.

**Pricing.** Open source is free. Cloud Free covers 25,000 spans a month. Pro is $19 a month for 100,000 spans, then $5 per 100,000.

### Braintrust

**What it is.** An evaluation and observability platform built around experiments, datasets, and scorers, with online scoring of production logs.

**Best for.** Teams whose releases run on experiments, who want production logs to feed datasets and scorers.

**Limits.** Each score is metered, so scoring all production traffic gets expensive. Self-hosting is Enterprise only. It is not a gateway.

**Pricing.** Starter is free with 1 GB, 10,000 scores, and 14-day retention. Pro is $249 a month with 5 GB, 50,000 scores, and 30-day retention, then $3 per GB and $1.50 per 1,000 scores. Users are unlimited.

### Datadog LLM Observability

**What it is.** Datadog's LLM and agent tracing product, now sold as Agent Observability, with evaluations, datasets, and experiments on every tier. It can be bought without other Datadog products.

**Best for.** Teams whose services already run on Datadog and want LLM traces linked to APM and infrastructure.

**Limits.** No self-hosting option is described, and longer retention is a paid add-on. See [Datadog LLM Observability alternatives](/blog/datadog-llm-observability-alternatives/) for the full breakdown.

**Pricing.** Free covers 40,000 LLM spans a month with 15-day retention. Pro starts at $160 a month on an annual commitment for 100,000 LLM spans ($200 month-to-month, $240 on-demand). Only LLM spans are billed, not tool or retrieval spans.

## Agnost AI

**What it is.** Agent monitoring for AI agents in production: chat agents, voice agents (through transcripts), copilots, and MCP servers. It is OpenTelemetry-native and has Python and TypeScript SDKs. It works with the OpenAI Agents SDK, Vercel AI SDK, LangChain and LangGraph, Mastra, Pydantic AI, Google ADK, and Agno, and groups traffic by user and session.

**How it differs from Helicone.** Helicone logs requests. Agnost judges whole conversations. It detects failures, user frustration, intents, tool-call errors, loops, and silent wrong answers. Its online evals are plain-language rules: once a conversation has been idle for 15 minutes, an LLM judges it in a batch and marks each assistant message pass or fail against every rule. There is no sampling. Alerts go to Slack or email. When an alert fires, Auto Improve opens a pull request with a fix tested against past runs, showing the replay results, for a person to review.

**Best for.** Teams that outgrew request logs and now need to know which conversations went wrong and why.

**Limits.** Agnost is not a gateway and does not track cost, so it does not replace Helicone's routing or spend dashboards. Keep a gateway (Portkey, LiteLLM, or OpenRouter) for those. Verdicts are pass or fail, not scores. Judging runs in batches, not in real time. It has no offline datasets or experiments, and it is not open source.

**Pricing.** Free covers 1,000 events a month with 7-day retention. Starter is $49 a month (10,000 events). Pro is $499 a month (1,000,000 events). Enterprise is custom and includes self-hosted VPC deployment, audit logs, and SLAs.

## How to choose

| If you used Helicone for… | Pick |
|---|---|
| Routing, fallbacks, and guardrails, managed | Portkey (ask about the Palo Alto roadmap) |
| A self-hosted gateway with spend caps | LiteLLM, plus an observability tool |
| One API key for many models | OpenRouter, with Broadcast to your observability tool |
| Request logs, cost, and prompts, open source | Langfuse |
| Open-source tracing plus automatic checks on every trace | Laminar |
| Low-cost open-source evals, including whole threads | Opik |
| Experiments and datasets | Braintrust |
| Everything else already runs on Datadog | Datadog LLM Observability |
| Finding which conversations failed and getting fixes as PRs | Agnost AI, next to a gateway |

Most teams leaving Helicone end up with two pieces: a gateway for routing and spend, and an observability or monitoring tool for traces and quality. Helicone combined the two, and few tools do both well.

## Migrating off Helicone

How hard the move is depends on how you integrated Helicone.

1. **Gateway (proxy) integration.** If your client's base URL points at `ai-gateway.helicone.ai` with a Helicone key, logging happens in the request path. To move, point the base URL at your new gateway (Portkey, LiteLLM, OpenRouter, or Langfuse's beta gateway), or straight back to the provider. Then turn on that gateway's logging or forwarding: Portkey logs natively, LiteLLM's callbacks and OpenRouter's Broadcast send traces to Langfuse, Datadog, or any OpenTelemetry collector. This is a config change, but test fallbacks, retries, and caching, since each gateway configures them differently.
2. **Async logging integration.** If you used Helicone's async logging through OpenLLMetry, your app already emits OpenTelemetry spans. Change the OTLP exporter endpoint and headers to the new backend. Langfuse, Laminar, Opik, Braintrust, Datadog, and Agnost AI all accept OpenTelemetry. Our [OpenTelemetry instrumentation guide](/blog/instrument-ai-agent-opentelemetry/) shows the setup.
3. **Proxy logs versus SDK or OTel traces.** A gateway only sees model calls. It doesn't see tool calls, retrieval, or how an agent's steps connect. If your app is now an agent, add OpenTelemetry or SDK instrumentation instead of relying only on gateway logs. You get full traces and you aren't tied to one proxy.
4. **Export what you need first.** Prompts, datasets, and any history you must keep are the slowest to rebuild. Export them before you switch traffic.
5. **Run both for a week.** Send the same traffic to Helicone and the candidate and compare what each records. The [agent monitoring pilot checklist](/blog/agent-monitoring-pilot-checklist/) lists what to check.

## Related reading

- [Best AI agent monitoring tools (2026)](/blog/best-ai-agent-monitoring-tools-2026/)
- [LangSmith alternatives for agent evals and monitoring](/blog/langsmith-alternatives/)
- [Datadog LLM Observability alternatives (2026)](/blog/datadog-llm-observability-alternatives/)
- [Langfuse alternatives for user and conversation analytics](/blog/langfuse-alternatives-user-analytics/)
- [AgentOps alternatives (2026)](/blog/agentops-alternatives/)
- [AI agent monitoring pricing guide](/blog/ai-agent-monitoring-pricing-cost-guide/)

## Frequently asked questions

### Is Helicone shutting down?

Helicone has not announced a shutdown. Mintlify announced on March 3, 2026 that it acquired Helicone and that "Helicone will continue operating in maintenance mode—security updates, bug fixes, and new models will keep shipping," adding that it will "work closely with every customer to support a smooth migration to another platform." Helicone's own site now says "Helicone is now part of Experiential Labs. Details to come soon." The pricing page still lists Hobby, Pro and Team plans.

### What are the best Helicone alternatives?

If you want a gateway, look at Portkey, LiteLLM, or OpenRouter. If you want open-source observability, look at Langfuse (which also has a gateway in public beta), Opik, or Laminar. Braintrust fits teams that run releases on experiments. Datadog fits teams already on Datadog APM. Agnost AI fits teams that need whole production conversations judged and failures turned into pull requests, though it does not track cost.

### What is the best open-source alternative to Helicone?

Langfuse is the closest: MIT-licensed except its ee folders, free to self-host, with tracing, prompts, evals and a gateway in public beta. LiteLLM's proxy is MIT outside its enterprise directory and is the most widely used open-source gateway. Portkey's gateway repo is MIT. Opik and Laminar are Apache-2.0 observability platforms, not gateways.

### Is there a Helicone alternative that is also an AI gateway?

Yes. Portkey, LiteLLM, and OpenRouter all route requests across providers. Portkey and OpenRouter log requests themselves, and OpenRouter's Broadcast feature forwards traces to tools like Langfuse, Datadog, or any OpenTelemetry collector. LiteLLM's proxy sends logs to external tools such as Langfuse, Datadog, and OpenTelemetry. Langfuse's own gateway is in public beta. Note that Palo Alto Networks completed its acquisition of Portkey on May 29, 2026.

### How do I migrate off Helicone?

If you used Helicone's gateway, change your client's base URL and API key to the new gateway, or back to the provider, and add an observability tool. If you used async logging through OpenLLMetry, you already emit OpenTelemetry, so you change the exporter endpoint and headers. Run both for a week and compare what each records before you cut over.

### Which Helicone alternative tracks LLM cost?

Portkey, LiteLLM, OpenRouter, Langfuse, Laminar, Opik, Braintrust, and Datadog all record token usage, and most show cost. Agnost AI does not track cost, so if spend dashboards were the main reason you used Helicone, pair Agnost with a gateway or pick one of the others.

### Is Helicone still a good choice in 2026?

For new projects, we would not start on it. Its acquirer says it is in maintenance mode and that it will help customers migrate elsewhere. If you already run it, it still works and still gets fixes, so plan a migration on your own schedule rather than in a rush.
