New plans are on the way.
MARK-17 is moving to simple monthly and yearly plans. Prices will be published here when checkout opens. There is no checkout yet, and nothing to buy today.
Join the Launch ListThe first 25 MARK-17 launch seats get launch pricing. No payment today.
The first 25 MARK-17 launch seats get launch pricing.
We’re keeping the first group small. That way we can help each person get set up, and listen to what they need.
People on the list who tick the box hear first when checkout opens. No payment today.
A short form on this site: your name, your email, and whether you want launch updates. No payment today.
What happens next
- 1 You join the list: your name, your email, the version you want, and one box to tick.
- 2 If you ticked the box, we email you launch updates and early access details. If you did not, we do not email you.
- 3 When checkout opens, people who ticked the box hear first. The first 25 MARK-17 launch seats get launch pricing.
Judge the product before you talk to anyone.
What MARK-17 does is published here, with screenshots of the real application.
What it does
Every capability, organised by area, with screenshots of the real application.
See the capabilities →What it tests
Every domain and test suite, and what each one actually measures.
Browse the tests →How it works
The full workflow from describing a job to sharing the evidence.
See the workflow →Questions.
Can I buy or download MARK-17 today?
Not yet. There is no checkout and no public download today. MARK-17 is moving to simple monthly and yearly plans. Join the MARK-17 Launch List and tick the box to hear first when checkout opens. No payment today.
What will MARK-17 cost?
Prices will be published here when checkout opens. The first 25 MARK-17 launch seats get launch pricing, and people on the MARK-17 Launch List who tick the box hear first.
Is my benchmark data uploaded anywhere?
No. MARK-17 is a desktop application and your benchmark data stays on your machine. There is no cloud sync of your results. If you benchmark a cloud AI provider, your test prompts are sent to that provider as part of normal API usage — the results still come back and are stored locally.
Which AI providers can I connect?
MARK-17 ships with a managed built-in llama.cpp runtime, so you can benchmark local models without installing anything else first. You can also connect Ollama, LM Studio, or LocalAI if you already use them. Cloud providers include OpenAI, Anthropic, Google Gemini, xAI (Grok), Groq, DeepSeek, Mistral, and Together AI, plus any custom OpenAI-compatible endpoint. Cloud providers use your own API keys.
What operating systems are supported?
MARK-17 is available first for Windows 10 and 11 (64-bit). macOS and Linux versions are coming soon. They are not available today, and there is no date yet. Join the MARK-17 Launch List, choose your platform, and tick the box to hear when your version is ready.
Do I need a GPU?
No. You do not need a GPU to run MARK-17, and you do not need one to benchmark cloud models — that work happens on the provider’s hardware. Local models can run CPU-only as long as you have enough system RAM; they are simply slower. A GPU with enough VRAM makes local benchmarking substantially faster and lets you run larger local models. MARK-17 reports how each model fits your hardware, so you can see what your machine will actually manage before committing to anything.
Is there a refund policy?
Yes. The full terms are set out in our Refund Policy.