diff --git a/CHANGES b/CHANGES
index 139ed6443..c8e390b8c 100644
--- a/CHANGES
+++ b/CHANGES
@@ -3,6 +3,7 @@ Features:
  - DS GX: Toon shading
  - DS GX: Clear depth
  - DS GX: Position test
+ - DS GX: Vector result
 Bugfixes:
  - DS GX: Fix vertex texture transformation (fixes mgba.io/i/702)
  - DS GX: Automatically normalize winding culling calculations (fixes mgba.io/i/699)
diff --git a/include/mgba/internal/ds/io.h b/include/mgba/internal/ds/io.h
index f4c5b7121..8355e8cfe 100644
--- a/include/mgba/internal/ds/io.h
+++ b/include/mgba/internal/ds/io.h
@@ -555,7 +555,6 @@ enum DS9IORegisters {
 	DS9_REG_VECMTX_RESULT_0F = 0x69E,
 	DS9_REG_VECMTX_RESULT_10 = 0x6A0,
 	DS9_REG_VECMTX_RESULT_11 = 0x6A2,
-	DS9_REG_VECMTX_RESULT_12 = 0x6A4,
 
 	DS9_REG_MAX = 0x106E,
 
diff --git a/src/ds/gx.c b/src/ds/gx.c
index c1d7c1392..4ac267600 100644
--- a/src/ds/gx.c
+++ b/src/ds/gx.c
@@ -162,6 +162,27 @@ static void _updateClipMatrix(struct DSGX* gx) {
 	gx->p->memory.io9[DS9_REG_CLIPMTX_RESULT_1F >> 1] = gx->clipMatrix.m[15] >> 16;
 }
 
+static void _updateVecMatrix(struct DSGX* gx) {
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_00 >> 1] = gx->vecMatrix.m[0];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_01 >> 1] = gx->vecMatrix.m[0] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_02 >> 1] = gx->vecMatrix.m[1];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_03 >> 1] = gx->vecMatrix.m[1] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_04 >> 1] = gx->vecMatrix.m[2];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_05 >> 1] = gx->vecMatrix.m[2] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_06 >> 1] = gx->vecMatrix.m[4];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_07 >> 1] = gx->vecMatrix.m[4] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_08 >> 1] = gx->vecMatrix.m[5];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_09 >> 1] = gx->vecMatrix.m[5] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_0A >> 1] = gx->vecMatrix.m[6];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_0B >> 1] = gx->vecMatrix.m[6] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_0C >> 1] = gx->vecMatrix.m[8];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_0D >> 1] = gx->vecMatrix.m[8] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_0E >> 1] = gx->vecMatrix.m[9];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_0F >> 1] = gx->vecMatrix.m[9] >> 16;
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_10 >> 1] = gx->vecMatrix.m[10];
+	gx->p->memory.io9[DS9_REG_VECMTX_RESULT_11 >> 1] = gx->vecMatrix.m[10] >> 16;
+}
+
 static inline int32_t _lerp(int32_t x0, int32_t x1, int32_t q, int64_t r) {
 	int64_t x = x1 - x0;
 	x *= q;
@@ -743,6 +764,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				gx->pvMatrixPointer -= offset;
 				memcpy(&gx->vecMatrix, &gx->vecMatrixStack[gx->pvMatrixPointer & 0x1F], sizeof(gx->vecMatrix));
 				memcpy(&gx->posMatrix, &gx->posMatrixStack[gx->pvMatrixPointer & 0x1F], sizeof(gx->posMatrix));
+				_updateVecMatrix(gx);
 				break;
 			case 3:
 				mLOG(DS_GX, STUB, "Unimplemented GX MTX_POP mode");
@@ -780,6 +802,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 			case 2:
 				memcpy(&gx->vecMatrix, &gx->vecMatrixStack[offset], sizeof(gx->vecMatrix));
 				memcpy(&gx->posMatrix, &gx->posMatrixStack[offset], sizeof(gx->posMatrix));
+				_updateVecMatrix(gx);
 				break;
 			case 3:
 				mLOG(DS_GX, STUB, "Unimplemented GX MTX_RESTORE mode");
@@ -795,6 +818,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				break;
 			case 2:
 				DSGXMtxIdentity(&gx->vecMatrix);
+				_updateVecMatrix(gx);
 				// Fall through
 			case 1:
 				DSGXMtxIdentity(&gx->posMatrix);
@@ -820,6 +844,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				break;
 			case 2:
 				memcpy(&gx->vecMatrix, &m, sizeof(gx->vecMatrix));
+				_updateVecMatrix(gx);
 				// Fall through
 			case 1:
 				memcpy(&gx->posMatrix, &m, sizeof(gx->posMatrix));
@@ -850,6 +875,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				break;
 			case 2:
 				memcpy(&gx->vecMatrix, &m, sizeof(gx->vecMatrix));
+				_updateVecMatrix(gx);
 				// Fall through
 			case 1:
 				memcpy(&gx->posMatrix, &m, sizeof(gx->posMatrix));
@@ -876,6 +902,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				break;
 			case 2:
 				DSGXMtxMultiply(&gx->vecMatrix, &m, &gx->vecMatrix);
+				_updateVecMatrix(gx);
 				// Fall through
 			case 1:
 				DSGXMtxMultiply(&gx->posMatrix, &m, &gx->posMatrix);
@@ -906,6 +933,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				break;
 			case 2:
 				DSGXMtxMultiply(&gx->vecMatrix, &m, &gx->vecMatrix);
+				_updateVecMatrix(gx);
 				// Fall through
 			case 1:
 				DSGXMtxMultiply(&gx->posMatrix, &m, &gx->posMatrix);
@@ -939,6 +967,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				break;
 			case 2:
 				DSGXMtxMultiply(&gx->vecMatrix, &m, &gx->vecMatrix);
+				_updateVecMatrix(gx);
 				// Fall through
 			case 1:
 				DSGXMtxMultiply(&gx->posMatrix, &m, &gx->posMatrix);
@@ -970,6 +999,7 @@ static void _fifoRun(struct mTiming* timing, void* context, uint32_t cyclesLate)
 				break;
 			case 2:
 				DSGXMtxTranslate(&gx->vecMatrix, m);
+				_updateVecMatrix(gx);
 				// Fall through
 			case 1:
 				DSGXMtxTranslate(&gx->posMatrix, m);
diff --git a/src/ds/io.c b/src/ds/io.c
index 4c44a0461..b58ec3216 100644
--- a/src/ds/io.c
+++ b/src/ds/io.c
@@ -604,7 +604,7 @@ void DS9IOWrite(struct DS* ds, uint32_t address, uint16_t value) {
 		value = ds->video.renderer->writeVideoRegister(ds->video.renderer, address, value);
 	} else if ((address >= DS9_REG_B_DISPCNT_LO && address <= DS9_REG_B_BLDY) || address == DS9_REG_B_MASTER_BRIGHT) {
 		value = ds->video.renderer->writeVideoRegister(ds->video.renderer, address, value);
-	} else if ((address >= DS9_REG_RDLINES_COUNT && address <= DS9_REG_VECMTX_RESULT_12) || address == DS9_REG_DISP3DCNT) {
+	} else if ((address >= DS9_REG_RDLINES_COUNT && address <= DS9_REG_VECMTX_RESULT_11) || address == DS9_REG_DISP3DCNT) {
 		value = DSGXWriteRegister(&ds->gx, address, value);
 	} else {
 		uint16_t oldValue;
@@ -719,7 +719,7 @@ void DS9IOWrite8(struct DS* ds, uint32_t address, uint8_t value) {
 }
 
 void DS9IOWrite32(struct DS* ds, uint32_t address, uint32_t value) {
-	if ((address >= DS9_REG_RDLINES_COUNT && address <= DS9_REG_VECMTX_RESULT_12) || address == DS9_REG_DISP3DCNT) {
+	if ((address >= DS9_REG_RDLINES_COUNT && address <= DS9_REG_VECMTX_RESULT_11) || address == DS9_REG_DISP3DCNT) {
 		value = DSGXWriteRegister32(&ds->gx, address, value);
 	} else {
 		switch (address) {
@@ -900,6 +900,24 @@ uint16_t DS9IORead(struct DS* ds, uint32_t address) {
 	case DS9_REG_CLIPMTX_RESULT_1D:
 	case DS9_REG_CLIPMTX_RESULT_1E:
 	case DS9_REG_CLIPMTX_RESULT_1F:
+	case DS9_REG_VECMTX_RESULT_00:
+	case DS9_REG_VECMTX_RESULT_01:
+	case DS9_REG_VECMTX_RESULT_02:
+	case DS9_REG_VECMTX_RESULT_03:
+	case DS9_REG_VECMTX_RESULT_04:
+	case DS9_REG_VECMTX_RESULT_05:
+	case DS9_REG_VECMTX_RESULT_06:
+	case DS9_REG_VECMTX_RESULT_07:
+	case DS9_REG_VECMTX_RESULT_08:
+	case DS9_REG_VECMTX_RESULT_09:
+	case DS9_REG_VECMTX_RESULT_0A:
+	case DS9_REG_VECMTX_RESULT_0B:
+	case DS9_REG_VECMTX_RESULT_0C:
+	case DS9_REG_VECMTX_RESULT_0D:
+	case DS9_REG_VECMTX_RESULT_0E:
+	case DS9_REG_VECMTX_RESULT_0F:
+	case DS9_REG_VECMTX_RESULT_10:
+	case DS9_REG_VECMTX_RESULT_11:
 	case DS9_REG_B_BG0CNT:
 	case DS9_REG_B_BG1CNT:
 	case DS9_REG_B_BG2CNT: