Dashboard
Cost, speed, and call volume on one screen
Single OpenRouter bill + 4 key metrics + today's alerts + trends by model and time
Beta — 7-day free Starter trialGet started →From writing to monitoring and rollback
No code changesRun prompts your way
Edit on the web and it ships to production in 5 minutes
Cost, quality, and access — all on one screen
- LLM call costs included in your subscription
- Change on the web with no code edits
- Auto-deployed to production within 5 minutes
Product screens
Here's what it looks like
Prompts
A list, not code
Per-environment separation + model and version control + weekly call-volume trend
Per-environment + call-volume trend
Editor
{{variable}} auto-detection + instant run
System message · model selection · fill variables and run, all on one screen
Auto variable detection
Version control
Compare line-by-line changes + roll back in one click
Separate deploys for production, staging, and dev + system prompt change comparison
Prompt change comparison
Cost analysis
Analysis by budget, time, and model
Monthly budget + daily trend + by time of day + model comparison, all on one screen
Monthly budget progress
0prompts running in production·0teams using PromptOps
01
Sign up and write a prompt
Sign up in a minute — the web editor auto-detects {{variables}} and runs them instantly
02
Install the SDK in one line
npm install @promptops/sdk — one line in your existing code picks up new versions automatically
03
Ship to production in 5 minutes
Just save in the web app — cost, evaluation, and rollback all live on one dashboard
Core features
Cost, quality, and access in one place
Measure cost and latency per prompt in real time
See latency and cost for every prompt at a glance — switch to a faster, cheaper model without touching your code
- Real-time cost dashboard
- Side-by-side model comparison
Daily cost by model
Last 24 hours · support-reply
- GPT-5.5$42.3+8%Avg. response 1.2s
- Claude Opus 4.7$18.1-3%Avg. response 0.8s
- Gemini 3.1 Pro$8.2-12%Avg. response 0.5s
Est. monthly cost $2,052−$1,680 by switching to Gemini 3.1 Pro
Score Korean prompt quality automatically
Add just 50 sample inputs and AI grades them across 7 criteria (accuracy, clarity, tone, safety, format, relevance, helpfulness) — verify quality before you ship
- Automatic scoring across 7 criteria
- Per-version score trends + alerts
Korean auto-evaluation
support-reply v13 · 50 LLM-graded samples
- Accuracy4.6
- Clarity4.3
- Tone & politeness4.8
- Safety5.0
- Format compliance4.1
- Relevance4.5
- Helpfulness4.4
+0.3 vs. previous v120 quality regressions
Role-based access and change history
Separate edit, approve, and deploy permissions by role — every change is logged with who, when, and what
- Separate edit, approve, and deploy roles
- Automatic change logging
Version history
support-reply- v13“Reply kindly, in three sentences or fewer…”PROD
- v12"Reply in the following format…"Revert
- v11"An answer to the user's question…"Revert
- v10"Initial version"
13 versions · auto-savedLive in production
Who it's for
Every role on your team can use it together
PM · Planning
Edit prompts on the web without knowing code — changes go live instantly
Developers
Use new versions automatically with one SDK line, no code changes
Content · Marketing
Manage tone and voice in one place — verify quality with automatic Korean eval
Support team
A/B test response templates — verify answer quality before deploying to production
How it works
From writing to monitoring and rollback
in one flow
The moment you edit on the web, the SDK picks up the new version and cost, latency, and eval scores accumulate automatically
If quality drops, roll back in one click
- ✎RunningEdit prompts on the web0 lines of code · one ⌘S
- ⇋SDK uses the new version instantlyNo redeploy · live in 5 minutes
- ◐Cost and latency measured automaticallyPer-prompt KPIs on one screen
- ↩Roll back in one click if quality dropsEval drops? Recover to v7 instantly
And more
Auto-extract prompts from your code
One CLI command scans your existing code for prompts — migrate them all into PromptOps at once
Human review · verify retrieval sources
When LLM scoring is ambiguous, people label cases by hand and verify RAG source grounding (Team).
BYOK and subscription credits
OpenRouter call costs are included in your subscription — bring your own API key to use it with no credit deduction
Use it with confidence
Korea region (Seoul)
TLS in transit · encryption at rest
Your API keys, envelope-encrypted
Workspace-isolated storage
365-day change history
Start with the plan that fits your team
Individual
Free
For evaluation / personal experiments
₩0/ mo
- ~250 LLM calls/month included
- 10 prompts · 1 user
- BYOK registration (no credit deduction)
Recommended
Growth
Starter
For solo PMs and developers
₩49,000/ mo
- ~4,400 LLM calls/month included
- 50 prompts · 90-day version retention
- Full Korean auto-evaluation
- Per-environment deploys (test/production)
Collaboration
Team
3-person team (+₩30,000/seat from the 4th)
₩99,000/ mo
- ~8,800 LLM calls/month included (shared by 3)
- 500 prompts · 365-day version retention
- Compare 3+ models at once · custom domain
- 24-hour email support
FAQ
Frequently asked questions
What's the difference between Free and Starter?+
Free is for evaluation with up to 250 calls/month, while Starter is built for real operations with up to 4,400 LLM calls/month
What happens to costs if I use my own API key?+
Register your own OpenAI · Anthropic · Google API key and LLM call costs are billed directly to your account — available even on the Free plan
How do I move prompts from my existing code?+
One CLI command auto-extracts them (`npx @promptops/cli scan`) — it recognizes yaml · json · Markdown · and even strings inside code
What does it cost to add team members?+
The Team plan includes 3 people for ₩99,000/month, with each additional person from the 4th costing ₩30,000
How accurate is the Korean evaluation?+
AI grades automatically across 7 criteria (accuracy, clarity, tone, safety, format, relevance, helpfulness) — and you can also have a person review ambiguous answers