Build your agent. Your way.

Four ways to play. Pick your build, pick how you pay.

Game Modes Switch anytime

Solo

Your prompt · Your key

Skip training. Paste your own prompt, plug in your own OpenAI / Anthropic / Gemini key, and ship in 90 seconds.

₹999 / month per agent
  • Paste your own base prompt
  • WhatsApp, Slack, Webhook & Email triggers
  • RAG, multi-modal, tools by toggle
  • Your API key handles all LLM cost

For: agencies, developers, prompt engineers

🛡

Co-op

We train · Your key

Brainam writes and refines the prompt with you in the Training Room. Once it's polished, switch the agent to your own LLM key to lock in low per-turn cost.

₹999 / month + training chips
  • Brainam-written, evolving base prompt
  • Practice Mode & sample queries
  • Your API key takes over after training
  • Best long-term LLM economics

For: polished SMBs who want training help + cost control

🎮

Quick Match

Your prompt · Our chips

Have a prompt ready but don't want a subscription commitment? Try Brainam with your own prompt and pay only for what you use from a chip top-up.

from ₹499 chip top-up
  • Paste your own base prompt
  • No subscription · pay-as-you-go
  • Test multiple models without setup
  • Upgrade to Solo + BYOK anytime

For: curious power users, trial / quick demos

Switch between modes anytime — your trained prompts and capabilities survive every swap.
Same agent infrastructure underneath all four builds.

Chip Top-ups Lifetime validity

Starter

₹499

1.5M chips

Try it out. Enough for a few weeks of moderate usage.

₹0.33 per 1,000 chips

Scale

₹1,499

6M chips

Steady volume across multiple agents.

₹0.25 per 1,000 chips
Best Value

Power User

₹1,999

10M chips

Cheapest per-chip rate. Bulk-buy and forget.

₹0.20 per 1,000 chips

1 chip = 1 token · Never expire · Payments via Razorpay UPI / cards / netbanking

BYOK · Bring Your Own Keys

Already have an OpenAI / Anthropic / Gemini API key? Run any individual agent on your own keys and skip chip top-ups entirely for that agent. Pick monthly or pay once for lifetime access.

Monthly

₹999/ month

Auto-renews · Cancel anytime

  • Runs on your own LLM API keys
  • Zero chip deduction for that agent
  • Unlimited chats · only your provider bill
  • Covers training AND production turns
  • Switch back to chip-pay anytime
  • One subscription = one agent

Lifetime

₹4,999once

One-time payment · Forever access

  • Everything in Monthly
  • Pays off after about 5 months
  • No renewals · No expiry
  • Transferable across LLM providers
  • Best for agents you plan to keep live for years

How chips work No upload size caps

1 chip ≈ 1 token.
You pay only for what the agent actually uses on each turn.

Standard usage burns the fewest chips. Premium features burn more — but never put a hard cap on what you can upload, ask, or analyse.

Standard usage low chip burn

💬

Basic chat

Quick replies with fast models (Haiku, GPT-4o-mini, Gemini Flash).

📄

Documents

Read PDF / DOC / TXT — no page limit, no file-size cap.

🧠

Short-term memory

Conversation context carried across turns automatically.

📱

WhatsApp & webhooks

Plain text incoming, plain text outgoing. Same chip cost as basic chat.

💎 Premium features cost more chips per turn

Pro models

Claude Sonnet, GPT-4o, Gemini Pro — sharper reasoning, longer outputs.

📚

RAG · Knowledge retrieval

Search across your uploaded data — no chunk cap, no embedding cap.

📷

Multi-Modal · Vision

Image read, OCR, image generation — no image-size limit, no resolution cap.

🎙

Voice & Video

Audio transcription, text-to-speech, voice agents — no clip-length cap.

💭

Deep Thinking

Extended reasoning before reply — for harder, multi-step problems.

🔧

Live tool calls

CRM lookups, Slack actions, web search — the agent uses tools when it needs to.

Why do premium features cost more chips?

Premium models think harder, retrieve more context, and process larger inputs — images, audio, long documents. Each turn uses more tokens, and chips track tokens 1:1.

The trade-off: we don't cap upload size or input length. A 100-page PDF, a 5 MB image, a 10-minute audio clip — all work on a single chip-pay turn. You only burn what the agent actually processes.