💰 BYOK Cost Calculator · How much will you really pay?

Answer 3 questions. Get your transparent total monthly budget (Alpha subscription + AI APIs) in 30 seconds.

💡 Why BYOK = cheaper than classic SaaS

AI analysis SaaS apps (Simply Wall St, Koyfin AI, Quartr Pro...) charge 5-10× the raw cost of the API they use in the backend. Why? Because they have to absorb:

• Their cloud infrastructure (proxies, load balancers, monitoring): 30-40% of price
• Their customer support and ops team: 20-30%
• Their commercial margin (they're a company, not a charity): 30-50%
• Churn risk (they get paid even when you don't use it)

With BYOK, you pay AI directly to Anthropic / OpenAI / Google, at their exact billing price. Alpha is just a prompt orchestration layer ($9.99/month). Zero markup on AI.

Concrete comparison for 50 analyses/month:

• Equivalent SaaS: ~$30-50/mo (plan including AI)
• Alpha BYOK: ~$15/mo ($9.99 Alpha + ~$5 API at cost)
• Alpha BYOK with free tier: $9.99/mo (GitHub Models / Cerebras free)

Typical savings: 50-80% over 12 months. Bonus: if you don't use the app for 1 month, you only pay the Alpha subscription — API at $0.

🆓 The 5 free / near-free providers

You can use Alpha without spending a cent on AI:

1. GitHub Models — free access to GPT-4o, Claude, Llama, Mistral via a GitHub PAT. Rate limit: ~50 requests/day. Perfect for casual.
2. Cerebras — Llama 70B at ultra-fast inference (>2000 tok/s). Free tier ~1M tokens/day.
3. Mistral La Plateforme — Mistral Small free with reasonable rate limit. EU data, GDPR-friendly.
4. Google AI Studio — Gemini 2.5 Flash free with generous quota (1500 requests/day).
5. Groq — Llama 3.3 70B at ultra-fast inference. Free tier for personal use.

👉 Full guide: how to get a free API key in 5 minutes.

🛡️ How to cap your budget

Three complementary mechanisms guarantee you'll never pay more than expected:

1. Monthly cap inside Alpha — Settings → Advanced → "Monthly budget cap". Set a value ($ or €, e.g. $5/month). Alpha tracks your usage in real time and blocks new calls once the cap is hit. Visual notification in the top bar.

2. Per-analysis cap — Settings → Advanced → "Per-analysis budget cap". Set a threshold ($) above which the analysis is canceled before sending. Safety net against heavy analyses (10-K on Berkshire = ~30k input tokens).

3. Provider-side limits — each provider (Anthropic, OpenAI, Google) lets you set hard limits in your account. E.g. Anthropic Console → Settings → Spend Limits. Set a hard cap (e.g. $20/month). Beyond that, the provider itself blocks requests.

With these 3 cumulative safeguards, surprise risk is near zero. Recommendation: start with an Alpha cap at $5/month, observe your real usage for 2-3 weeks, adjust.

❓ FAQ