Comet ML
ML experiment tracking + LLM observability
Overview
Comet is an established ML experiment-tracking platform that has expanded into LLM observability via the Opik project. Used by enterprise ML teams to track training runs, hyperparameter sweeps and now LLM-app traces.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 20 May 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
Use cases
What you can produce with Comet ML
- Add a few lines of Python to a PyTorch or scikit-learn training script and automatically log metrics, hyperparameters, code versions, and artifacts for every run.
- Compare dozens of training runs side by side on interactive dashboards to spot which hyperparameter combination actually moved the validation curve.
- Register a trained model with lineage back to the exact experiment, dataset, and code commit that produced it before promoting it to production.
- Trace an LLM application end to end with Opik, inspecting each prompt, retrieval step, tool call, and response in a RAG or agent pipeline.
- Score LLM outputs at scale with Opik's automated evaluations, including hallucination and relevance checks, and gate deployments on the results in CI.
- Monitor production LLM traffic on Opik 2.0 dashboards to track cost, latency, and quality regressions over time.
- Self-host the open-source Opik stack so prompts and traces never leave your infrastructure, then upgrade to the managed platform only if you need it.
ASEAN Perspective
Comet ML in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
Comet ML is a well-established MLOps platform for experiment tracking, model registry, production monitoring and, increasingly, LLM evaluation/observability via its open-source Opik project. It is a credible Weights & Biases alternative with strong logging, comparison and reproducibility features, plus self-hosted/on-prem options for regulated teams.
It is aimed squarely at ML/data-science teams, so there is a learning curve and the value really shows at team scale. The free tier and OSS Opik lower the barrier for individuals. Docs and SDKs are good. For ASEAN ML teams it is globally available with self-host options, though no localised support. Solid, if not the flashiest, choice.
What people say
Comet is one of the longer-standing independent players in ML experiment tracking, and since late 2024 it has run a second act: Opik, its open-source LLM observability and evaluation platform, which has built a genuinely large GitHub following and gained custom dashboards plus cost and latency tooling in its 2.0 release. The company remains independent and actively developed, which matters in a category where several rivals have been acquired or drifted.
User sentiment on the core tracking product is positive but thin on the big review sites; it sits around 4.3/5 on G2 from a small pool of about a dozen reviews, so treat any aggregate rating with caution. What reviewers on G2 and Capterra consistently praise is how little boilerplate it takes to start logging, easy integration with mainstream frameworks, clear dashboards for comparing training runs and hyperparameter sweeps, and unusually responsive support, including direct Slack access. The free tier for individuals and free Pro access for verified academics earn frequent goodwill from students and researchers.
The complaints follow familiar lines. Enterprise licensing costs are described by some teams as prohibitive, and in head-to-head comparisons G2 reviewers tend to favour Weights & Biases on product support quality and roadmap direction, a sign Comet is the challenger rather than the default in many evaluations. Teams running both classic training and LLM apps also report friction from the Comet-plus-Opik split: effectively two products, two SDKs, and two billing lines, with metadata correlation between a training run and its LLM traces requiring extra work.
Comet fits ML teams that want mature, low-friction experiment tracking without building on a hyperscaler's stack, academics who can use it free, and LLM developers happy to self-host Opik as a standalone open-source tool. Teams whose workload is LLM-first from day one may prefer starting with Opik alone or an LLM-native observability platform.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Comet ML's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Comet ML unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Comet ML directly →
Spotted something out of date? Suggest an update →
More in Code & Dev Tools
