Pricing
Priced by the work.
Never by surprise.
Every plan includes the web app, the REST API and MCP. Each action shows its credit price before it runs, failed work gives its credits back, and questions about your research are free every day.
Plan estimator
How much research will you run?
Published price list
What things cost.
Credits are a stable work unit. The price of every paid action is published here and shown again before it runs, so token counts and provider prices never surprise you.
Research
- Standard research3 to 4 rounds and one cited synthesis
- Unavailable
- Deep researchMore research passes, focused follow-on investigations, and an included dossier
- Unavailable
- Rabbit holeOne focused child investigation
- Unavailable
- DossierA set of research deliverables
- Unavailable
- Research workflowEach step costs what its action costs. You see the most a run can cost before it starts
- Per step
Questions
- Ask your researchA cited answer across one report (up to 3) or several (up to 6). Charged for the work it took, never more than that
- Up to 3, or 6
- Ask a follow-up on a reportA source-grounded answer from one report
- Unavailable
- Follow-up with web searchThe same answer, plus a search for newer sources
- Unavailable
Always free
- Read or exportYour existing work stays free to use
- Free
Prepaid credits
Top up without upgrading.
Packs never expire while your subscription is active. Mini and Plus are for the Solo and Starter plans, and every plan can buy the other packs. Each pack shows what a credit costs.
Published pack prices. Checkout shows the current price before you pay.
- MiniSolo & Starter500credits for $15$0.030 per credit
- PlusSolo & Starter1,000credits for $29$0.029 per credit
- Top-upAll plans2,500credits for $120$0.048 per credit
- SmallAll plans5,000credits for $220$0.044 per credit
- MediumAll plans25,000credits for $950$0.038 per credit
- LargeAll plans100,000credits for $3,200$0.032 per credit
- BulkAll plans500,000credits for $14,000$0.028 per credit
Overage is off by default
If you turn it on, you set a monthly dollar cap. We warn you at 80 percent, and work stops at 100 percent.
Auto top-up is optional
Pick a pack, a balance threshold and a monthly limit. When your balance drops below the threshold we buy that pack with your saved card, never more than your limit in a month.
Powerful open source engines
Free in the open.
Licensed for closed code.
Every engine we release, starting with StateFrame, is free under the AGPL-3.0-only. When your own code stays closed, one commercial license from Farside Research LLC covers every released engine, priced by company size instead of seats.
- Solo1 person$9.99per monthor $99.99 per year
- Team2 to 10 people$99per monthor $990 per year
- Company11 to 50 people$390per monthor $3,900 per year
- Growth51 to 250 people$1,200per monthor $12,000 per year
- Scale251 to 1,000 people$3,000per monthor $30,000 per year
- Enterprise1,001+ peopleFrom $60,000per yearBy request
- Free to evaluate: the AGPL lets you run the full engine before you buy.
- Unlimited developers, machines, and deployments inside your company.
- Monthly by card, or yearly by card or invoice. Separate from your research plan, with no Farside account needed.
Questions
Nothing hidden in the footer.
What is a credit?
A credit is a stable work unit. Every paid action publishes its credit price before you run it, so provider pricing and token counts never surprise you.
Do unused credits roll over?
Yes. Unused monthly credits roll forward for one billing period, up to one month of your plan allowance. Prepaid packs never expire while the subscription is active.
What happens when I run out?
New paid work stops. Farside Research does not silently bill you. Reads and exports keep working. If you turned on auto top-up, it refills your balance with the pack you chose once it drops below your threshold, up to the monthly limit you set.
Do you charge extra for API or MCP access?
No. Every plan includes both.
What happens when a run fails?
Unused reserved credits are returned automatically. You do not have to ask.
Do you train on my research?
No. Your questions, sources, and reports are not used to train models.
Start with one question.
Pick a plan, ask your first question, and keep every source behind the answer.