Free for everyone during beta.

No card. No sign-up. No paid tier yet.

ManyMinds is in public beta. There’s no sign-up, no card, and no paid tier yet. Use the analyzer as much as you’d like — within reasonable rate limits — and tell us what’s working and what isn’t.

A paid tier will follow once we’ve heard from enough of you to know what’s worth charging for. We’ll give plenty of notice before anything changes; today’s free usage stays free.

What we’re thinking for the paid tier

(Subject to change based on your feedback.)

FeatureToday (beta)Future paid tier (preview)
Bias scoring engineHeuristic (regex + lexicon, 10 parameters)LLM-as-Judge (Gemini Flash US / DeepSeek V3 China — pick one)
Per-parameter reasoningNumbers + evidence snippetsNumbers + evidence snippets + LLM-written rationale
Bias-dial revisionRewrite the chosen response toward neutralized targets (watermarked)
Self-bias check (your own writing)Paste a draft article / thesis / memo, see how it reads vs LLMs
Sign-in requiredNoYes
Analyses per day5–20 (rate-limited)200
Analyses per month30–1002,000
Target LLMs per call2–42–7
Prompt cap (tokens)1,0006,000
User-output cap2,00024,000
Response cap per target8003,000
Open-weight providersAll 8 (US 4 / EU 1 / China 3)Same
Closed-API providersOpenAI, Anthropic, Google, Cohere, Grok Mini, Chinese flagships
Export (Markdown / PDF)YesYes
Copy + upload affordancesYes (coming this week)Yes
API accessYes (2,000 calls/month included)

Why we’ll eventually charge

A lot of the open-weight models we run today are technically free — Meta Llama, Qwen, Mistral, DeepSeek and others publish their weights for anyone to use. What isn’t free is the hardware those models actually run on. Every comparison you run on ManyMinds is a request to a serverless GPU somewhere, billed by the second by the cloud providers we use. The closed-API providers — OpenAI, Anthropic, Google, Cohere — bill us per token by name.

So when the paid tier arrives, the price will reflect the real cost of running models on real hardware, plus the cost of keeping the lights on (servers, monitoring, support, occasional security audits). We’d rather charge a fair monthly fee than show ads, sell your data, or compromise the neutrality we’re trying to build. If the paid tier never feels worth it for what you do, the free tier won’t disappear.

What should the paid tier look like?

Tell us which models matter to you, what caps would actually fit your work, and what would make this worth $X/month. Email us at hello@manyminds.in →