AI Infrastructure · Top 10
The 10 Best Prompt Management Tools in 2027
Direct answer
Langfuse is the best pick on our rule: its own pages name prompt versioning, evaluation, tracing and a free plan, and its Core plan is $29 per month. LangSmith is #2 and PromptLayer is #3. Langfuse is also the best value, as the lowest-priced pick on the chart that lists a free plan. Humanloop is not ranked because its own page says it is joining Anthropic and sunsetting the platform.
At a glance
| # | Pick | Price | Best for |
|---|---|---|---|
| 1 | Langfuse Prompt Management | Core is $29 per month and lists 100k units a month included, then $8 per 100k units; the page text read shows no monthly or yearly toggle (read 2026-11-08) | Teams that want prompt versioning, tracing and evals in one open-source tool with a free plan and a clearly stated $29 monthly start |
| 2 | LangSmith | Plus is $39 per seat per month then pay as you go; Developer is $0 per seat per month (read 2026-11-08) | Teams already on LangChain tools who want prompt iteration, evals and tracing together |
| 3 | PromptLayer Prompt Registry | Pro is $49 per month; Team is $500 per month; Pro adds pay-as-you-go at $0.003 per transaction (read 2026-11-08) | Teams where non-engineers edit and version prompts in a visual editor |
| 4 | Arize Phoenix | Phoenix is open source and free to run; Arize AX Pro, the managed platform, is $50 per month, with no billing toggle shown in the page text read (read 2026-11-08) | Teams that want to run tracing, evals and prompt iteration on their own infrastructure |
| 5 | Portkey Prompt Library | Production is $49 per month, with 100k recorded logs a month and $9 for each additional 100k requests (read 2026-11-08) | Teams that want prompt templates and versions attached to an AI gateway |
| 6 | Weights & Biases Weave | Pro starts at $60 per month, billed monthly; the pricing page lists W&B plans and does not price Weave alone (read 2026-11-08) | Teams already tracking experiments in Weights & Biases who want LLM tracing and evals alongside |
| 7 | Helicone | Pro is $79 per month with usage-based pricing on top and a 7-day free trial; Team is $799 per month (read 2026-11-08) | Teams that want request logging and cost tracking first, with prompts and datasets alongside |
| 8 | Braintrust | Pro is $249 per month; Starter is $0 per month (read 2026-11-08) | Teams that care most about evals and trace review, with prompts compared inside experiments |
| 9 | PromptHub Prompt Library | Not stated clearly: the pricing page has a Monthly or Yearly toggle labelled Yearly (Save 25%) and the page text lists both $12 and $9 for Pro, and $20 and $15 per user for Team (read 2026-11-08) | Small teams that want Git-style prompt versioning and a free way to start |
| 10 | Promptfoo | Not stated clearly: Community is free forever, and Enterprise and On-Premise are custom; no paid plan price is shown (read 2026-11-08) | Developers who want open-source prompt and model evals in their CI pipeline |
Pick by what you want
Best overall
Langfuse Prompt Management
Best value
Langfuse Prompt Management
Best for non-engineers editing prompts
PromptLayer Prompt Registry
Best for self-hosting an open-source tool
Arize Phoenix
Best for open-source prompt evals in CI
Promptfoo
The ranking
1
Langfuse Prompt Management
Answrs VerdictOpen-source prompt versioning with tracing and evals, a free plan and the lowest clear monthly price here.
Price: Core is $29 per month and lists 100k units a month included, then $8 per 100k units; the page text read shows no monthly or yearly toggle (read 2026-11-08)Prompt management: Langfuse describes prompt management as storing, versioning and retrieving prompts centrally, with labels, configuration and client-side caching in the SDKs; the pricing page lists Prompt Versioning and Prompt Release ManagementEvaluation and observability: The pricing page lists tracing, online and offline evaluation, datasets, experiments, LLM-as-judge evaluators and human annotation queuesFree plan: Hobby is free with no credit card, 50k units a month, 30 days of data access and 2 usersOpen source: The pricing page says Langfuse is open source and can be self-hosted for free; it also shows a Langfuse joins ClickHouse story link
Langfuse describes prompt management as storing, versioning and retrieving prompts centrally, with labels and SDK caching, and lists tracing, evaluation and datasets too. Hobby is free with 50k units a month, and Core is $29 per month. It says it is open source and shows a Langfuse joins ClickHouse story link.
Go ifTeams that want prompt versioning, tracing and evals in one open-source tool with a free plan and a clearly stated $29 monthly start
Skip ifYou need a flat per-seat price: Langfuse bills in usage units
Why #1 over #2: It scores on all five parts and has the lowest monthly price among the top-scoring picks.
2
LangSmith
Answrs VerdictPrompt versions, evals and tracing from LangChain, priced per seat plus usage.
Price: Plus is $39 per seat per month then pay as you go; Developer is $0 per seat per month (read 2026-11-08)Prompt management: LangChain says you can experiment with prompts in the Playground and compare outputs across different prompt versions or model providersEvaluation: LangSmith lists offline evaluation on curated datasets and online evaluation of production traffic, with human review queuesObservability: LangSmith lists agent tracing, monitoring with cost tracking, online LLM-as-judge evals and alerts via webhook and PagerDutyFree plan: Developer is $0 per seat per month for solo users, with up to 5k base traces a month, then pay-as-you-goHosting: Self-hosted and hybrid deployment options are listed under Enterprise; no open-source licence is stated for LangSmith on the pages read
LangSmith lets you experiment with prompts in the Playground and compare outputs across prompt versions, alongside offline and online evals and agent tracing. Developer is $0 per seat per month and Plus is $39 per seat per month, then pay as you go.
Go ifTeams already on LangChain tools who want prompt iteration, evals and tracing together
Skip ifYou want open source or a flat price: pricing is per seat plus usage
Why #2 over #3: It scores on all five parts, but its per-seat price is higher than Langfuse’s.
3
PromptLayer Prompt Registry
Answrs VerdictA visual prompt registry with versions, evals and traces, from $49 a month.
Price: Pro is $49 per month; Team is $500 per month; Pro adds pay-as-you-go at $0.003 per transaction (read 2026-11-08)Prompt management: PromptLayer says you can visually edit, version, test and deploy prompts without an application redeploy, and diff versions and roll back changesEvaluation and observability: PromptLayer lists dataset-backed tests, human review and automated graders, plus agent observability that connects production traces to prompt versionsFree plan: Free is $0 per month with 5 users, 2.5k requests a month and 10 promptsHosting: Pro and Team are cloud-hosted in the US; the pricing page lists self-hosted deployment only on Enterprise; no open-source claim was found on the pages read
PromptLayer says you can visually edit, version, test and deploy prompts, diff versions and roll back. Free is $0 per month with 2.5k requests, and Pro is $49 per month. Self-hosting is listed only on Enterprise.
Go ifTeams where non-engineers edit and version prompts in a visual editor
Skip ifYou want open source: self-hosting is Enterprise-only
Why #3 over #4: It scores on all five parts, but its monthly price is above Langfuse and LangSmith.
4
Arize Phoenix
Answrs VerdictSelf-hostable tracing, evals and prompt iteration, with a $50 managed tier.
Price: Phoenix is open source and free to run; Arize AX Pro, the managed platform, is $50 per month, with no billing toggle shown in the page text read (read 2026-11-08)Prompt management: The Arize AX pricing table lists Prompt serving and versioning, trace replay in the playground and multi-prompt comparison; Phoenix links to a Prompt IDEEvaluation and observability: Arize calls Phoenix an open-source, local-first platform for tracing, evaluation, experimentation and prompt iterationFree plan: AX Free lists 25k trace spans a month, 1 GB of ingestion and 15 days retentionOpen source: Phoenix is described as open source and self-hostable, and its page shows ELv2 licensed
Arize describes Phoenix as an open-source platform for tracing, evaluation, experimentation and prompt iteration, self-hostable and ELv2 licensed. Its managed AX Pro plan is $50 per month and lists prompt serving and versioning.
Go ifTeams that want to run tracing, evals and prompt iteration on their own infrastructure
Skip ifYou want one price for the open-source tool: paid plans belong to AX
Why #4 over #5: It scores on all five parts, but its paid price belongs to the separate AX product.
5
Portkey Prompt Library
Answrs VerdictPrompt templates with versioning on a gateway, from $49 a month.
Price: Production is $49 per month, with 100k recorded logs a month and $9 for each additional 100k requests (read 2026-11-08)Prompt management: Portkey lists prompt templates with a playground, versioning, variables and API endpoints; Developer allows 3 templates and Production lists unlimited templatesObservability: Portkey lists logs, traces, feedback, custom metadata and filters, with alerts on ProductionFree plan: Developer is free forever with 10k recorded logs a month; Portkey says it is not suitable for production workloadsOpen source: The pricing page shows an Open Source row with Host it Yourself and Deploy Locally; which plans it applies to is not stated clearly
Portkey lists prompt templates with versioning and variables on an AI gateway, plus logs and traces. Developer is free with 10k recorded logs, and Production is $49 per month. Evaluation is named only under Enterprise guardrails.
Go ifTeams that want prompt templates and versions attached to an AI gateway
Skip ifYou need evals: Portkey names evaluation templates only under Enterprise guardrails
Why #5 over #6: It scores four of five parts because evaluation is not named outside Enterprise, and wins the price tie-break over the other four-point picks.
6
Weights & Biases Weave
Answrs VerdictTracing and evals with a free plan and a $60 monthly start, but no named prompt registry.
Price: Pro starts at $60 per month, billed monthly; the pricing page lists W&B plans and does not price Weave alone (read 2026-11-08)Prompt management: The Weave docs list Experiment with prompts and models and Version your work as guides; they do not name a prompt registryEvaluation and observability: Weave is described as an observability and evaluation platform for tracing, evaluating and improving agents and LLM applicationsFree plan: Free is $0 per month and lists AI application evaluations, tracing and scorers, with 1 GB a month of Weave data ingestionOwnership: The docs sit on a CoreWeave Docs site and say CoreWeave Agent Lens is the successor to Weights & Biases Weave
Weave is described as an observability and evaluation platform for agents and LLM apps, with guides for experimenting with prompts and versioning your work. Free is $0 per month and Pro starts at $60 per month, billed monthly. The docs say CoreWeave Agent Lens is its successor.
Go ifTeams already tracking experiments in Weights & Biases who want LLM tracing and evals alongside
Skip ifYou want a dedicated prompt registry: the docs read do not name one
Why #6 over #7: It scores four of five parts, with no named prompt registry, and costs less than Helicone and Braintrust.
7
Helicone
Answrs VerdictOpen-source logging and gateway with prompts and datasets, from $79 a month plus usage.
Price: Pro is $79 per month with usage-based pricing on top and a 7-day free trial; Team is $799 per month (read 2026-11-08)Prompt management: The pricing table lists Prompts, Playground, Scores and Datasets under Prompts & TestingObservability: Helicone lists requests, sessions, user analytics, alerts and reports, plus a gateway with caching and rate limitsFree plan: Hobby is free with 10,000 requests a month, 1 GB storage and 1 seatOpen source and status: The pricing page says Helicone is open source; the homepage says Helicone is no longer part of Mintlify and that details are to come
Helicone lists prompts, a playground, scores, datasets, request analytics and a gateway. Hobby is free with 10,000 requests a month, and Pro is $79 per month plus usage. Its homepage says Helicone is no longer part of Mintlify.
Go ifTeams that want request logging and cost tracking first, with prompts and datasets alongside
Skip ifYou need evaluation named as a feature: the pricing table lists scores and datasets
Why #7 over #8: It scores four of five parts and costs less than Braintrust.
8
Braintrust
Answrs VerdictStrong evals and tracing, but no named prompt registry and a $249 paid start.
Price: Pro is $249 per month; Starter is $0 per month (read 2026-11-08)Evaluation: Braintrust lists experiments against versioned datasets, side-by-side comparison of prompts and models, and scoring by LLMs, code or humansPrompt management: The pages read mention a playground and prompt iteration with the Loop agent; a versioned prompt registry is not namedFree plan: Starter is free with 1 GB processed data, 10k scores and 14-day retention, with unlimited users and projectsHosting: Enterprise lists on-prem or hosted deployment; the pages read describe open source libraries and tools, not an open-source platform
Braintrust lists experiments against versioned datasets, side-by-side prompt and model comparison, and tracing. Starter is free with 10k scores, and Pro is $249 per month. A versioned prompt registry is not named on the pages read.
Go ifTeams that care most about evals and trace review, with prompts compared inside experiments
Skip ifYou want a prompt registry or a low monthly price
Why #8 over #9: It scores four of five parts but has the highest monthly price on the chart.
9
PromptHub Prompt Library
Answrs VerdictGit-style prompt versioning with a free plan, but an unclear paid price.
Price: Not stated clearly: the pricing page has a Monthly or Yearly toggle labelled Yearly (Save 25%) and the page text lists both $12 and $9 for Pro, and $20 and $15 per user for Team (read 2026-11-08)Prompt management: PromptHub describes Git-based versioning with a simple API and deploying prompts to environments through branchesEvaluation: PromptHub says you can evaluate prompts at scale in a simple UI and run pipelines of evaluators on every commit or mergeFree plan: Free has unlimited team members and unlimited public prompts, with no private promptsObservability: Request tracing is not named in the pages read; plans list monthly request limits
PromptHub describes Git-based prompt versioning, evaluation at scale and pipelines of evaluators on every commit. Free has unlimited public prompts but no private ones. The paid price is not stated clearly because the page lists both $12 and $9 for Pro.
Go ifSmall teams that want Git-style prompt versioning and a free way to start
Skip ifYou need production tracing or a clear price
Why #9 over #10: It scores three of five parts, with no tracing named and no clear paid price.
10
Promptfoo
Answrs VerdictOpen-source evals and red teaming, but no prompt registry or tracing named.
Price: Not stated clearly: Community is free forever, and Enterprise and On-Premise are custom; no paid plan price is shown (read 2026-11-08)Evaluation: Promptfoo lists Evaluations as a product for testing and evaluating prompts, models and RAG pipelines; the Community plan lists all LLM evaluation featuresFree plan: Community is free forever and includes red teaming at 10k probes a monthOpen source: The pricing page calls Community its open-source tool, run locally or self-hostedPrompt management: Prompt versioning and production tracing are not named on the pages read; the homepage centres on red teaming and security testing
Promptfoo’s Community plan is a free, open-source tool that lists all LLM evaluation features and red teaming at 10k probes a month. Enterprise is custom, no paid price is shown, and prompt versioning is not named on the pages read.
Go ifDevelopers who want open-source prompt and model evals in their CI pipeline
Skip ifYou need a prompt registry or production tracing
Related guides
If you are also guarding prompts against attack, our infrastructure guide covers prompt injection defence.
- How do you prevent prompt injection at the infrastructure layer in 2027? · Defending prompts at the infrastructure layer
How we ranked these
We gave each tool one point for each of five things its own pages name: prompt management or versioning; evaluation; observability or tracing; a free plan; and a cheapest paid plan with a clearly stated price. Ties go to the lower monthly price shown, picks with no clear price last, then alphabetical order. Best value is the lowest-priced pick on the chart that lists a free plan; that is Langfuse. All facts come from each maker’s own pages, read on 2026-11-08. The chart compares one metric: the cheapest paid plan in USD per month. Three picks are left off it: LangSmith is priced per seat, PromptHub’s price is not stated clearly, and Promptfoo shows no paid price. Humanloop is not ranked because its own page says it is joining Anthropic and sunsetting the platform. Vellum’s pricing page now describes a personal assistant platform, Agenta’s pages describe an agent workspace, and Latitude’s pages describe agent observability, with no prompt management named on the pages read, so none was ranked. Pezzo returned a not found page and Lilypad did not respond. Facts are taken from each maker's own published information; check the official site before you buy.
Questions people ask
What is the best prompt management tool?
Langfuse ranks first on our rule: its pages name prompt versioning, evaluation, tracing and a free plan, and its Core plan is $29 per month. LangSmith, PromptLayer and Arize Phoenix also score on all five parts.
Which tool is the best value?
Langfuse: it has the lowest monthly price on the chart, $29 per month for Core, among picks that list a free plan. PromptLayer and Portkey are next at $49 per month.
Which prompt management tools are open source?
Langfuse says it is open source and self-hostable, Arize says Phoenix is open source and self-hostable, Helicone says it is open source, and Promptfoo calls its Community tool open source. LangSmith, PromptLayer, Portkey, Braintrust and PromptHub do not state an open-source licence on the pages read.
What happened to Humanloop?
Humanloop’s own homepage says its team is joining Anthropic and that it is sunsetting the Humanloop platform, so we did not rank it.
Cheapest paid plan in USD per month (flat monthly platform price; per-seat, unclear and unpriced plans are left out) at a glance
Langfuse Prompt Management29 USD per month
PromptLayer Prompt Registry49 USD per month
Portkey Prompt Library49 USD per month
Arize Phoenix50 USD per month
Weights & Biases Weave60 USD per month
Helicone79 USD per month
Braintrust249 USD per month
Official sites
- Langfuse Prompt Management · langfuse.com
- LangSmith · langchain.com
- PromptLayer Prompt Registry · promptlayer.com
- Arize Phoenix · phoenix.arize.com
- Portkey Prompt Library · portkey.ai
- Weights & Biases Weave · weave-docs.wandb.ai
- Helicone · helicone.ai
- Braintrust · braintrust.dev
- PromptHub Prompt Library · prompthub.us
- Promptfoo · promptfoo.dev
Sources
- langfuse.com
- langchain.com
- langchain.com
- promptlayer.com
- promptlayer.com
- arize.com
- portkey.ai
- portkey.ai
- wandb.ai
- helicone.ai
- braintrust.dev
- prompthub.us
- promptfoo.dev
- humanloop.com
- vellum.ai
- agenta.ai
- latitude.so
Prices, specs and availability change; confirm on the official site before you buy.