SIMD-0650 proposes a Solana syscall opcode that returns the BN254 pairing result instead of a yes-or-no flag
The Solana Improvement Documents repository merged SIMD-0650 on 31 October. The proposal, written by authors at Anza, has the status Idea and extends SIMD-0302. It would add a new opcode, ALT_BN128_PAIRING_MAP, to the existing sol_alt_bn128_group_op syscall that programs use for elliptic-curve operations on the BN254 curve.
Today the pairing operation, ALT_BN128_PAIRING, mirrors Ethereum's precompile. It takes a list of point pairs, computes the product of their pairings and reports only whether that product equals the identity element, as a 32-byte flag. That is enough for Groth16 proof verification, which can be arranged as a single identity check, but it never exposes the resulting element of the target group. The new opcode computes the same product and writes the target group element itself: twelve 32-byte field coefficients, 384 bytes in total.
The proposal lists what this would allow. A program could compare the pairing product with a value computed off-chain and stored on-chain instead of re-deriving it each time; split an expensive verification across several transactions by combining partial products later; and use the pairing output directly as key material for identity-based or time-lock encryption. It would also bring BN254 in line with the BLS12-381 syscalls proposed in SIMD-0388, whose pairing already returns the target group element.
The design keeps everything else unchanged. The new opcode takes the next free number, 7, after opcodes 4 to 6 allocated by SIMD-0302. Its input format and validation are identical to the existing pairing operation, so callers can switch opcodes without re-encoding data, and input length must be a multiple of 192 bytes as defined in SIMD-0334. The little-endian flag from SIMD-0284 applies to both inputs and output. Compute units are charged per pair, so the transaction budget limits the work.

What it means
The change is small at the syscall level but widens what zero-knowledge and cryptography programs can do on Solana. As an Idea-stage proposal it still needs a feature gate and implementation before it could activate on the network.