These two solve different problems, and only one is taking new customers. Helicone, a gateway and request logger, has been in maintenance mode since March 2026 with cloud sign-ups disabled; Weave traces, evaluates and monitors agents and is the practical choice for a new deployment. Existing Helicone users should read this page as migration guidance.[3][6]
The difference between Helicone and W&B Weave starts with status: Helicone has been in maintenance mode since March 2026 with new cloud sign-ups disabled, while Weave is actively developed inside CoreWeave Forge. Helicone is an LLM request logger and gateway that by its own docs is not an evaluation framework; Weave is a tracing, evaluation and monitoring toolkit. For a new deployment Weave is usually the practical choice of the two; existing Helicone users should plan a migration.
| Decision factor | Helicone | W&B Weave |
|---|---|---|
| Status | Maintenance mode since March 2026; new cloud sign-ups disabled; ownership unclear since 19 October 2026 | Active; sold inside CoreWeave Forge since 30 September 2026 |
| What it is | LLM request logger and AI gateway; not an evaluation framework, per its docs | Tracing, evaluation, monitoring and guardrails toolkit |
| Agent tracing | Sessions grouped by request headers; no agent span model | Threads, nested calls and an Agents view for OTel GenAI spans |
| Evaluations | Scores posted from external tools; Python evaluators turned off October 2026 | Evaluation API, LLM-judge and code scorers, production monitors |
| Gateway | Yes: OpenAI-compatible gateway for 100+ providers | None documented |
| Self-hosting | Free, Apache-2.0, via Docker or Helm | Kubernetes with a Weave licence and a W&B Platform install |
| Entry pricing | Hobby free, Pro $79 a month; new sign-ups disabled | Forge Free: 1 GB of Weave ingestion a month; Pro from $60 a month |
| Capability | Helicone | W&B Weave | 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 | ✓ | ✓ |
| Prompt versioning and playground | ✓ | ✓ | — |
| Act on findings | |||
| Runtime controls on live agent actions | Partial | Partial | ✓ |
| Gateway or proxy in the request path | ✓ | Not stated | — |
| Licence and deployment | |||
| Open source | ✓ | Partial | Partial |
| Self-hosting | ✓ | ✓ | Partial |
Mintlify announced its acquisition of Helicone on 3 March 2026 and moved the product into maintenance mode: patches, bug fixes and new model support only, with help for customers to move to other platforms. Cloud sign-ups have shown 'Sign ups are disabled' since the same day, and the changelog has no feature entries after November 2025. On 19 October 2026 Helicone's site began saying it is no longer part of Mintlify, with details to come and no new owner named. In May 2026 an AWS account lock took its dashboard and log processing down for about four days, though the proxy stayed up.
Weave is in a different position. CoreWeave completed its acquisition of Weights & Biases in May 2025 and on 30 September 2026 launched CoreWeave Forge, which now sells Weave alongside W&B Models and new tools such as Agent Lens. Weave's SDK has shipped releases every month through 2026. Its packaging is still changing after the Forge launch, but it is being developed.
For a new deployment, then, Weave is usually the practical choice of the two. Helicone remains relevant mainly to teams already running it, especially self-hosted.
Even setting status aside, the products do different jobs. Helicone sits in the request path: its OpenAI-compatible gateway routes calls to more than 100 providers with failover, caching, rate limits and per-request cost, and logs each call. Its docs say plainly that Helicone does not run evaluations; scores from external tools can be posted to requests. Python evaluators were turned off in October 2026, and its Experiments feature was removed in 2025.
Weave is out of the request path. A decorator or OpenTelemetry instrumentation traces your functions, and Weave adds what Helicone lacks: an Evaluation API over datasets, LLM-judge and code scorers, monitors on production traces and annotation queues. It documents no gateway. Both offer versioned prompts and a playground. For agents, Helicone groups requests into sessions using request headers, a path-based hierarchy rather than spans; Weave shows OTel GenAI spans as agents, turns, LLM calls and tool calls.
On cost and run-time checks, Helicone's gateway gives per-request cost from its model registry, cost alerts with thresholds, and can block prompt injection and jailbreak attempts, which its docs say works with OpenAI models only. Weave calculates token costs for supported providers from its Python SDK, and its guardrails are scorers called from Python code that can block or modify inputs and outputs. Neither documents approvals on an agent's tool calls.
How hard the move is depends on how you integrated Helicone. If you used its async logging, that is built on OpenLLMetry instrumentation, and Weave reads OpenLLMetry attributes over OTLP, which gives you a starting point. If you used the proxy as your integration, Weave has no equivalent: you will need separate instrumentation for tracing, and a separate gateway if you rely on Helicone's routing, failover or caching.
Self-hosted Helicone is Apache-2.0 and will keep running, so there is no forced deadline. Self-hosting Weave needs a Weave-enabled licence and a W&B Platform install, so teams moving for that reason should price the licence first. Note that Helicone's separate Rust gateway repository is GPL-3.0 and has not been updated since November 2025.
As of 20 October 2026, Helicone's pricing page lists Hobby free with 10K requests, Pro at $79 a month and Team at $799 a month, but new cloud sign-ups are disabled, so new customers cannot start on any plan. CoreWeave prices Weave through Forge: Forge Free includes 1 GB of Weave ingestion a month and up to 5 model seats, and Forge Pro starts at $60 a month with 1.5 GB for teams under 50 employees, with extra ingestion listed at $0.10 per MB. Helicone counts requests; Weave counts bytes. See Helicone's pricing and CoreWeave Forge 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 gateway can block prompt injection for OpenAI models only, and Weave's guardrails check content from Python code; neither holds an agent's action for approval. Prefactor is also a migration option for Helicone users, but it has no gateway, no prompt playground or prompt registry, and its platform is hosted. 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.