Skip to content
Skip to content

Fees and Limits

Balchemy does not have subscription plans, pricing tiers, or locked features. Every user gets the same access to the platform. This page explains the fees that do apply, where they come from, and what "enterprise" actually means.


How Balchemy makes money

Balchemy earns revenue through a small fee on each swap execution. There is no monthly fee, no per-seat charge, and no gating of features behind a paywall. If you trade, a small percentage goes to Balchemy. If you do not trade, there is no cost.


Trading fees

Standard fee

The default trading fee is 1% of the swap amount. This applies to all bots and external agents unless the bot is configured with enterprise policy mode.

Enterprise fee

Bots configured with policyMode: "enterprise" pay a reduced fee of 0.5%. This is a bot-level configuration, not a subscription or account upgrade. You set it on the bot itself.

The enterprise policyMode is not a tier — it is a field on the Bot schema (policyMode: 'standard' | 'enterprise'). It affects fee rates and some advanced behavior like auto-analysis of token queries. You configure it in Studio when creating or editing a bot.

Fee calculation

Fees are calculated in basis points and applied using integer arithmetic (BigInt) to avoid floating-point errors across all token decimals:

  • Solana tokens (9 decimals), EVM tokens (18 decimals), and USDC (6 decimals) are all handled correctly
  • Net amount = gross amount − fee
  • The fee is always floored (rounded down), so you never pay more than the stated rate
  • Fees are recorded in an append-only ledger with the order reference, amount, currency, and type

Solana fees (Jupiter)

On Solana, fees are collected through Jupiter's feeAccount mechanism. The fee is deducted from the output token before the swap result is delivered to your wallet. You see the net amount in your wallet — the fee never leaves your custody and then comes back; it is collected at the protocol routing level.

EVM fees (Base and Ethereum)

On EVM chains, fees are deducted before the swap executes. The flow is:

  1. Your input token amount is received
  2. The platform fee is deducted from the input
  3. The remaining net amount goes into the swap
  4. The swap output reaches your wallet

The distributeFee() function handles fee distribution to the platform treasury and any partner payouts.


Partner payouts

If your bot or agent was onboarded through a partner or referral program, a portion of the platform fee may be distributed to the partner. This is handled automatically by the trading engine and does not increase your total fee — the split comes out of Balchemy's share, not yours.


Referral commissions

When a referred user executes a trade, a referral commission is calculated from the platform fee and sent to the referrer as a fire-and-forget operation after the trade completes. The commission is capped and processed asynchronously — it does not delay your trade or affect your balance.

The referral commission is sourced from Balchemy's fee share, not from an additional charge on the trader.


MCP scopes

MCP access (connecting external AI agents or Claude Desktop to Balchemy) is free. The three scopes — read, trade, manage — have no additional cost. You pay trading fees only when an actual swap executes.

ScopeCostWhat it enables
readFreeMarket data, portfolio, research, logs
tradeFree (fees apply on swaps)Trading commands, approvals, execution
manageFree (requires claim + step-up)Key rotation, scope updates, control plane

AI usage quotas

Balchemy enforces daily and monthly AI usage quotas at the platform level to protect system stability. These are rate limits, not paywalls.

When you hit a quota:

  • The request returns an error indicating the limit has been reached
  • The quota resets on the daily or monthly boundary
  • Quotas are enforced per user, not per bot

Quotas are set server-side and are not currently user-configurable. If you are building an integration that generates high AI request volume, use the granular MCP tools (market data, wallet reads, etc.) rather than ask_bot or trade_command for every operation — they bypass the AI layer entirely and do not count toward AI quotas.


Rate limits

MCP calls are subject to rate limiting when MCP_RATE_LIMIT_ENABLED is active on the server. The platform also enforces Redis-backed per-user rate limits at the application layer.

When you exceed a rate limit you receive 429 Too Many Requests. Wait and retry — there is no penalty for hitting a rate limit.


What enterprise policyMode is (and is not)

"Enterprise" in Balchemy refers to a configuration field on a bot, not a subscription tier.

  • It is: policyMode: 'enterprise' on the Bot schema — a flag that reduces your fee rate from 1% to 0.5% and enables some advanced message-handler behavior (like auto-analysis of detected token queries)
  • It is not: a plan, a subscription, a Stripe product, or a feature gate
  • How you get it: you set it when creating or editing a bot in Studio, or via the bot API

There is no application process, no approval step, and no extra cost to use enterprise policyMode. It is a configuration choice.


Summary table

ItemValue
Monthly subscriptionNone
Standard trading fee1% per swap
Enterprise policyMode trading fee0.5% per swap
MCP scopesFree
Feature gatingNone — all features available to all users
Referral commission sourceBalchemy's fee share (no extra charge to trader)
Partner payout sourceBalchemy's fee share (no extra charge to trader)
Rate limitingYes (server-configurable, per user)
AI usage quotasYes (daily + monthly, per user)

Connection lost. Retrying...