Multi-Token Flash

Borrow multiple different tokens in a single atomic transaction. Essential for cross-token arbitrage strategies.

Transaction Pattern

Multi-token flash uses a nested sandwich — borrows are stacked, then repaid in reverse order:

text

┌──────────────────────────────────────────────┐
│            Single Atomic Transaction         │
│                                              │
│  IX 1:  begin_flash(SOL, 1000)               │
│  IX 2:  begin_flash(USDC, 50000)             │
│                                              │
│  IX 3:  your_arb_ix(SOL → DEX_A)            │
│  IX 4:  your_arb_ix(USDC → DEX_B)           │
│                                              │
│  IX 5:  end_flash(USDC)                      │
│  IX 6:  end_flash(SOL)                       │
│                                              │
│  All or nothing — atomic execution           │
└──────────────────────────────────────────────┘

Code Examples

const sig = await flash.borrowMulti({
  loans: [
    { token: 'SOL', amount: 1000 },
    { token: 'USDC', amount: 50000 },
  ],
  onFunds: async (ixs) => {
    ixs.push(arbSolIx);
    ixs.push(arbUsdcIx);
    return ixs;
  },
});

PDA Isolation

Each token gets its own PDA — ["flash", payer, token_mint]. This prevents cross-token collisions and allows independent routing for each borrow.

Limits

Constraint	Value	Reason
Max tokens per TX	4	Transaction size limit (1232 bytes)
Each token independently routed	✅	Each uses its own source/fallback
Partial repay	❌ All or nothing	Atomicity guarantee

💡 Tip

Keep multi-token borrows to 2-3 tokens max. Each borrow adds ~200 bytes and ~5,000 CU to the transaction.