How we make this list: we compare pricing pages, product documentation and public user reviews. Ratings come from G2, Capterra or Trustpilot, with the date we checked them.
Prompt engineering tools help teams write, version, test, and monitor the prompts that drive large language model applications, from drafting in a playground to running scored evaluations and watching live traffic. Prices below are third-party reported and checked 14 Nov 2026. Confirm the live price before you buy.
PromptLayer is the top pick for most teams that want non-engineers to edit prompts safely, because it pairs a visual prompt editor with versioning, evaluations, and release controls at a low entry price. LangSmith is the strongest alternative for teams already built on LangChain and LangGraph, where deep tracing and agent deployment matter more than a simple prompt editor.
Key takeaways
- PromptLayer leads for mixed teams where product and domain staff own prompt wording, with Pro reported at 49 dollars per month.
- LangSmith leads for LangChain and LangGraph teams, with Plus reported at 39 dollars per seat per month.
- Promptfoo is the best free starting point for developers who want version-controlled evaluations in code, with a free open source Community edition.
- Helicone is now in maintenance mode after its March 2026 acquisition by Mintlify, so new buyers should weigh self-hosting or a supported alternative.
- HoneyHive and Promptfoo have no self-serve paid tier with a published price.
How we picked
We picked tools that cover prompt versioning, testing, or live monitoring, and that a buyer can still sign up for or self-host in November 2026. We checked each vendor website to confirm it loads, and we checked third-party pricing trackers citing vendor pricing pages checked in September, October, and November 2026. We favoured tools with a free plan or a free open source edition, clear evaluation features, and support for more than one model provider. Prices are third-party figures because vendor pages change often.
Comparison table
| Tool | Best for | Starting price | Free plan |
| PromptLayer | Non-engineers editing prompts | 49 dollars a month, Pro | Yes, 2,500 requests a month |
| LangSmith | LangChain and LangGraph teams | 39 dollars a seat a month, Plus | Yes, Developer, 1 seat |
| Promptfoo | Code-based evals and red teaming | Free, Community; paid by quote | Yes, open source Community |
| Weights and Biases Weave | Teams already using Weights and Biases | 60 dollars a month, Pro | Yes, 1 GB ingestion a month |
| Helicone | Fast proxy-based cost tracking | 79 dollars a month, Pro | Yes, Hobby, 10,000 requests a month |
| Portkey | Production AI gateway with prompt templates | 49 dollars a month, Production | Yes, Developer, 10,000 logs a month |
| Agenta | Open source prompt playground and evals | 29 dollars a month, Starter | Yes, Hobby cloud and free self-hosted |
| HoneyHive | Small teams wanting evals on a free plan | Quote, Enterprise; Developer free | Yes, Developer, 10,000 events a month |
A small team should choose PromptLayer or Promptfoo first, depending on who edits the prompts. PromptLayer fits when product managers, support leads, or subject experts need to change wording without a code release. Promptfoo fits when engineers own the prompts and want tests in the same repository as the code. Both have a usable free tier, so a team can prove value before paying. Start with versioning and a small evaluation set, then add tracing once prompts reach production.
Prompt engineering tools cost from free to about 799 dollars per month for a published mid-tier plan, before usage charges. Entry paid plans run from 29 dollars per month for Agenta Starter to 79 dollars per month for Helicone Pro, with PromptLayer Pro and Portkey Production both reported at 49 dollars per month. LangSmith Plus is reported at 39 dollars per seat per month, so a ten-person team starts at 390 dollars per month before trace usage. The headline price is rarely the full bill, because usage charges sit on top of the base plan.
Most production teams need both, but they do not need to buy both on day one. Prompt management stores versions, controls who can publish, and lets a team roll back a bad change. Evaluation scores prompt outputs against test cases, so a team can tell whether a wording change helped or hurt before it reaches users. A team that only edits prompts in a shared document needs management first, while a team that already ships prompts through code needs evaluation first.
Check four things before you switch prompt tools: export, access control, evaluation history, and model coverage. Export matters because prompts are business logic, and a tool that locks versions inside a proprietary format creates a painful move later. Access control matters when non-engineers edit prompts, because a team needs approval steps and a clear audit trail.
Use the decision table below to match your team shape to a starting tool. Run the same ten prompts and the same evaluation set through two finalists and compare the scores.
| Team type and workflow | Start with | Why it fits |
| Mixed team, non-engineers edit prompts | PromptLayer | Visual editor and low entry price |
| LangChain or LangGraph team | LangSmith | Native tracing, evals, and deployment |
| Engineering-led team, prompts in Git | Promptfoo | YAML evals, red teaming, no seat cost |
| Machine learning team on Weights and Biases | Weave | Tracing and evals in the existing platform |
| Team needing cost visibility this week | Helicone or Portkey | Proxy or gateway cost tracking |
| Small team wanting one open source workspace | Agenta | Playground, evals, and free self-hosting |
| Small team wanting generous free evals | HoneyHive | Free plan with online evals |
| Production team needing guardrails | Portkey | Gateway, fallbacks, and prompt templates |
Quick comparison
| Tool | Best for | Starting price | Free plan |
|---|
 PromptLayer | Non-engineers editing prompts with versioning and release controls | 49 dollars per month for Pro, third-party reported, 14 Nov 2026 | Yes, free plan with 5 users, 2,500 requests per month, and 10 prompts |
 LangSmith | LangChain and LangGraph teams that need tracing, evals, and agent deployment | 39 dollars per seat per month for Plus, third-party reported, 14 Nov 2026 | Yes, Developer plan at no cost for 1 seat with 5,000 base traces per month |
 Promptfoo | Developers who want code-based prompt evals and red teaming in CI | Free for Community; Enterprise and On-Premise by quote | Yes, Community edition is free forever under the MIT licence |
 Weights and Biases Weave | Machine learning teams already using Weights and Biases for experiment tracking | 60 dollars per month for Pro, third-party reported, 14 Nov 2026 | Yes, free plan with 5 seats, 5 GB storage per month, and 1 GB Weave ingestion per month |
 Helicone | Teams that want fast proxy-based cost and request tracking | 79 dollars per month for Pro, third-party reported, 14 Nov 2026 | Yes, Hobby plan at no cost with 10,000 requests per month |
 Portkey | Production teams that need an AI gateway with prompt templates and guardrails | 49 dollars per month for Production, third-party reported, 14 Nov 2026 | Yes, Developer plan at no cost with 10,000 recorded logs per month |
 Agenta | Small teams that want an open source playground, evals, and tracing in one workspace | 29 dollars per month for Starter, third-party reported, 14 Nov 2026 | Yes, Hobby cloud plan at no cost and a free self-hosted open source edition |
 HoneyHive | Small teams that want online evals and human review on a free plan | Quote for Enterprise; Developer free | Yes, Developer plan at no cost with 10,000 events per month and up to 5 users |
Is your product missing or out of date?
We review this list regularly. Vendors can request a listing, send updated pricing or ask about featured placement.
#1
PromptLayer
Best for: Non-engineers editing prompts with versioning and release controls
Visit promptlayer.com
Starting price49 dollars per month for Pro, third-party reported, 14 Nov 2026
Free planYes, free plan with 5 users, 2,500 requests per month, and 10 prompts
PromptLayer lets teams version, test, deploy, and monitor prompts through a visual editor. Product, support, and domain staff can change wording without waiting for an engineer, while release labels route a set share of traffic to a new version. The free plan covers 2,500 requests per month with a 10-prompt cap. Pro is reported at 49 dollars per month with overage at 0.003 dollars per transaction. Price conflict: one September 2026 read of the vendor pricing page showed pay-as-you-go pricing with no fixed monthly price, while October and November 2026 trackers list Pro at 49 dollars per month.
What we like- Visual editor for non-engineers, versioning with release labels, evaluations and tracing
Watch out for- Price jumps from 49 dollars per month for Pro to 500 dollars per month for Team, and approvals and self-hosting sit on Enterprise
#2
LangSmith
Best for: LangChain and LangGraph teams that need tracing, evals, and agent deployment
Visit langchain.com
Starting price39 dollars per seat per month for Plus, third-party reported, 14 Nov 2026
Free planYes, Developer plan at no cost for 1 seat with 5,000 base traces per month
LangSmith is the hosted platform from LangChain for building, monitoring, and evaluating language model applications. It traces chains, agents, and tool calls, scores outputs against datasets, and supports deployment for long-running agents. The Developer plan is free for one seat with 5,000 base traces per month and 14-day retention. Plus is reported at 39 dollars per seat per month with 10,000 base traces included. Retention conflict: older sources describe extended retention as 400 days, while September and October 2026 sources describe it as 180 days for new traces.
What we like- Deep tracing for LangChain and LangGraph, strong evaluation and dataset tools, deployment in one platform
Watch out for- The free Developer plan allows only one seat, so a second teammate forces a move to Plus, and usage billing for traces, compute, and retention can push the bill well above the seat price
#3
Promptfoo
Best for: Developers who want code-based prompt evals and red teaming in CI
Visit promptfoo.dev
Starting priceFree for Community; Enterprise and On-Premise by quote
Free planYes, Community edition is free forever under the MIT licence
Promptfoo is an open source command-line tool for testing and evaluating language model applications and for red teaming them against security risks. Teams write test cases in YAML, run them against any provider, and compare outputs side by side. The Community edition is free forever under the MIT licence. OpenAI announced its acquisition of Promptfoo on 9 March 2026, and both sides stated that the project stays open source. Enterprise plans add team sharing and managed deployment, with no public price listed. Limit conflict: the vendor pricing page lists red teaming at 10,000 probes per month for Community, while the CLI source in version 0.125.1 sets the cap at 100,000 probes.
What we like- Open source CLI with YAML test suites, works with many providers, strong red teaming, runs in CI
Watch out for- No hosted team dashboard in the free edition, and dashboards and single sign-on need Enterprise
#4
Weights and Biases Weave
Best for: Machine learning teams already using Weights and Biases for experiment tracking
Visit wandb.ai
Starting price60 dollars per month for Pro, third-party reported, 14 Nov 2026
Free planYes, free plan with 5 seats, 5 GB storage per month, and 1 GB Weave ingestion per month
Weave is the language model evaluation and observability product from Weights and Biases, now sold through CoreWeave Forge after CoreWeave completed its acquisition in May 2025. It traces agent runs, scores outputs, and monitors production behaviour alongside model experiments. The free plan includes 1 GB of Weave ingestion per month. Pro starts at 60 dollars per month flat for up to 10 seats, with 1.5 GB of ingestion, then 0.10 dollars per MB after that. Price conflict: one older listing still shows paid plans from 20 dollars per month, while pricing pages checked in September, October, and November 2026 list Pro from 60 dollars per month.
What we like- Agent-native tracing, evaluation leaderboards, and guardrail scorers, Apache 2.0 SDK
Watch out for- Weave has no standalone plan, and ingestion overage at 0.10 dollars per MB on Pro can grow fast
#5
Helicone
Best for: Teams that want fast proxy-based cost and request tracking
Visit helicone.ai
Starting price79 dollars per month for Pro, third-party reported, 14 Nov 2026
Free planYes, Hobby plan at no cost with 10,000 requests per month
Helicone is an open source observability platform and AI gateway that sits between an application and model providers as a proxy. Changing the base URL gives a team request logs, token counts, latency, errors, and cost tracking across providers such as OpenAI, Anthropic, and Google Gemini. The Hobby plan is free with 10,000 requests per month, 1 GB storage, one seat, and 7-day retention. Pro is reported at 79 dollars per month with unlimited seats and one-month retention. Price conflict: older guides list Pro at 20 dollars per month, while pricing pages checked in September and October 2026 list Pro at 79 dollars per month.
What we like- One-line proxy setup, request logging with token counts and costs, semantic caching, open source self-hosted edition
Watch out for- Mintlify acquired Helicone on 3 March 2026 and the hosted product is now in maintenance mode, and new hosted signups are disabled
#6
Portkey
Best for: Production teams that need an AI gateway with prompt templates and guardrails
Visit portkey.ai
Starting price49 dollars per month for Production, third-party reported, 14 Nov 2026
Free planYes, Developer plan at no cost with 10,000 recorded logs per month
Portkey is an AI gateway and observability control plane for teams running language model and agent applications in production. It routes requests across providers, applies guardrails, caches responses, and tracks cost and latency in one dashboard. The Developer plan is free with 10,000 recorded logs per month and 3 prompt templates. Production is reported at 49 dollars per month with 100,000 recorded logs, 30-day retention, and unlimited prompt templates. Extra logs cost 9 dollars per additional 100,000 requests up to 3 million per month. The gateway core is open source under the MIT licence.
What we like- Unified gateway for more than 1,600 models, automatic fallbacks and retries, prompt templates, guardrails, caching
Watch out for- Log overage at 9 dollars per extra 100,000 logs can add up fast, and Palo Alto Networks acquired Portkey in May 2026
#7
Agenta
Best for: Small teams that want an open source playground, evals, and tracing in one workspace
Visit agenta.ai
Starting price29 dollars per month for Starter, third-party reported, 14 Nov 2026
Free planYes, Hobby cloud plan at no cost and a free self-hosted open source edition
Agenta is an open source LLMOps platform from Berlin that combines a prompt playground, an evaluation harness, and request tracing in one workspace. Teams can compare prompt variants across models, run evaluations, and trace requests through OpenTelemetry. The Hobby cloud plan is free with signup and daily credits, one project, and two members. Starter is reported at 29 dollars per month with 2,900 credits per month. Business is reported at 299 dollars per month with 32,000 credits. Price conflict: older listings show Pro at 49 dollars per month and Business at 399 dollars per month, while October and November 2026 sources list Starter at 29 dollars per month.
What we like- Side-by-side playground with model comparison, 15 or more built-in evaluators, OpenTelemetry tracing, no per-seat charge
Watch out for- Cloud billing moved to credits in Agenta 2.0, released in July 2026, so credits can run out faster than a trace allowance suggests
#8
HoneyHive
Best for: Small teams that want online evals and human review on a free plan
Visit honeyhive.ai
Starting priceQuote for Enterprise; Developer free
Free planYes, Developer plan at no cost with 10,000 events per month and up to 5 users
HoneyHive pairs OpenTelemetry tracing with evaluations, experiments, production monitoring, and human review. Teams can score live traffic, sample outputs for annotation, and block a release in CI when evaluation scores drop. The Developer plan is free with 10,000 events per month, 30-day retention, and up to 5 users. Enterprise is custom priced with custom limits and retention and a dedicated technical account manager. A production team should get a written Enterprise quote before committing, because there is no published paid price to budget against.
What we like- OpenTelemetry tracing with online evals, CI regression gates, annotation queues, and drift detection
Watch out for- There is no self-serve paid tier, so the step after free is a sales quote, and every span counts toward the 10,000-event cap
Frequently asked questions
What is a prompt engineering tool
A prompt engineering tool is software that helps a team write, store, version, test, and monitor the prompts sent to a large language model. It replaces scattered documents and hard-coded strings with a shared prompt registry, an editor, evaluation scores, and live traces. A good tool shows which prompt version is live, who changed it, and how it scored in tests.
How do I choose between PromptLayer and LangSmith
Choose PromptLayer when non-engineers own prompt wording and need a visual editor with release controls at a low entry price. Choose LangSmith when the team already builds with LangChain or LangGraph and needs deep tracing, dataset evaluations, and agent deployment. PromptLayer is simpler for mixed teams, while LangSmith is stronger for engineering-led agent work.
Are there free prompt engineering tools
Yes, every tool in this guide has a free plan or a free open source edition. Promptfoo Community and the Agenta self-hosted edition are free under open source licences with no seat cost. PromptLayer, LangSmith, Weave, Helicone, Portkey, Agenta Cloud, and HoneyHive all offer a free hosted tier with monthly caps. Free plans suit prototypes and small teams, but check the caps first.
Can I self-host a prompt engineering tool
Yes, several tools in this guide support self-hosting. Promptfoo runs locally as an open source CLI, Agenta offers a free self-hosted open source edition, Helicone offers an open source self-hosted edition under Apache 2.0, and Portkey offers an open source gateway under the MIT licence. LangSmith and PromptLayer self-hosting is Enterprise only, and HoneyHive self-hosting needs an Enterprise plan.
How do I test a prompt before it goes live
Test a prompt by running it against a fixed set of test cases and scoring the outputs before release. Start with ten to twenty cases that cover normal requests, edge cases, and known failure modes. Score each output with exact checks, model-based judges, and human review. Compare the new version against the current live version on the same set, and only promote it when the score improves or stays level.
Sources we checked (16)
- https://agentopsguide.com/prompt-management-tools/
- https://promptlayer.com
- https://evalsreview.com/pricing/langsmith
- https://www.langchain.com/langsmith
- https://www.aievals.co/tools/promptfoo
- https://promptfoo.dev
- https://panoply.io/tools/weights-and-biases/
- https://wandb.ai
- https://panoply.io/tools/helicone/
- https://www.helicone.ai
- https://agentsindex.ai/blog/litellm-vs-portkey-vs-helicone-ai-gateway
- https://portkey.ai
- https://panoply.io/tools/agenta/
- https://agenta.ai
- https://panoply.io/tools/honeyhive/
- https://www.honeyhive.ai
Is your product missing or out of date?
We review this list regularly. Vendors can request a listing, send updated pricing or ask about featured placement.
Shihab Mia · Founder, Rankite
Shihab has 5+ years in B2B and SaaS SEO. He writes Rankite's software buying guides from pricing pages, documentation and public user reviews. More about Shihab.