diff --git a/chacha20/chacha_loong64.go b/chacha20/chacha_loong64.go
new file mode 100644
index 0000000..9679f9e
--- /dev/null
+++ b/chacha20/chacha_loong64.go
@@ -0,0 +1,22 @@
+// Copyright 2026 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+//go:build gc && !purego
+
+package chacha20
+
+import "golang.org/x/sys/cpu"
+
+const bufSize = 256
+
+//go:noescape
+func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32)
+
+func (c *Cipher) xorKeyStreamBlocks(dst, src []byte) {
+	if cpu.Loong64.HasLSX {
+		xorKeyStreamVX(dst, src, &c.key, &c.nonce, &c.counter)
+	} else {
+		c.xorKeyStreamBlocksGeneric(dst, src)
+	}
+}
diff --git a/chacha20/chacha_loong64.s b/chacha20/chacha_loong64.s
new file mode 100644
index 0000000..e377b43
--- /dev/null
+++ b/chacha20/chacha_loong64.s
@@ -0,0 +1,342 @@
+// Copyright 2026 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// derived from chacha_arm64.s
+
+//go:build gc && !purego
+
+#include "textflag.h"
+
+DATA	·constants+0x00(SB)/4, $0x61707865
+DATA	·constants+0x04(SB)/4, $0x3320646e
+DATA	·constants+0x08(SB)/4, $0x79622d32
+DATA	·constants+0x0c(SB)/4, $0x6b206574
+GLOBL	·constants(SB), NOPTR|RODATA, $32
+
+DATA	·incRotMatrix+0x00(SB)/4, $0x00000000
+DATA	·incRotMatrix+0x04(SB)/4, $0x00000001
+DATA	·incRotMatrix+0x08(SB)/4, $0x00000002
+DATA	·incRotMatrix+0x0c(SB)/4, $0x00000003
+GLOBL	·incRotMatrix(SB), NOPTR|RODATA, $32
+
+#define NUM_ROUNDS 10
+
+// func xorKeyStreamVX(dst, src []byte, key *[8]uint32, nonce *[3]uint32, counter *uint32)
+TEXT ·xorKeyStreamVX(SB), NOSPLIT, $0
+	MOVV	dst+0(FP), R4
+	MOVV	src+24(FP), R5
+	MOVV	src_len+32(FP), R6
+	MOVV	key+48(FP), R7
+	MOVV	nonce+56(FP), R8
+	MOVV	counter+64(FP), R9
+
+	MOVV	$·constants(SB), R10
+	MOVV	$·incRotMatrix(SB), R11
+
+	MOVW	(R9), R12
+
+loop:
+	MOVV	$NUM_ROUNDS, R15
+	// load 4-32bit data from incRotMatrix added to counter
+	VMOVQ	(R11), V30
+
+	// load contants
+	VMOVQ	(R10), V0.W4
+	VMOVQ	4(R10), V1.W4
+	VMOVQ	8(R10), V2.W4
+	VMOVQ	12(R10), V3.W4
+
+	// load keys
+	VMOVQ	(R7), V4.W4
+	VMOVQ	4(R7), V5.W4
+	VMOVQ	8(R7), V6.W4
+	VMOVQ	12(R7), V7.W4
+	VMOVQ	16(R7), V8.W4
+	VMOVQ	20(R7), V9.W4
+	VMOVQ	24(R7), V10.W4
+	VMOVQ	28(R7), V11.W4
+
+	// load counter + nonce
+	VMOVQ	(R9), V12.W4
+
+	VMOVQ	(R8), V13.W4
+	VMOVQ	4(R8), V14.W4
+	VMOVQ	8(R8), V15.W4
+
+	// update counter
+	VADDW	V30, V12, V12
+
+chacha:
+	// V0..V3 += V4..V7
+	// V12..V15 <<<= ((V12..V15 XOR V0..V3), 16)
+	VADDW	V0, V4, V0
+	VADDW	V1, V5, V1
+	VADDW	V2, V6, V2
+	VADDW	V3, V7, V3
+	VXORV	V12, V0, V12
+	VXORV	V13, V1, V13
+	VXORV	V14, V2, V14
+	VXORV	V15, V3, V15
+	VROTRW	$16, V12, V12
+	VROTRW	$16, V13, V13
+	VROTRW	$16, V14, V14
+	VROTRW	$16, V15, V15
+
+	// V8..V11 += V12..V15
+	// V4..V7 <<<= ((V4..V7 XOR V8..V11), 12)
+	VADDW	V8, V12, V8
+	VADDW	V9, V13, V9
+	VADDW	V10, V14, V10
+	VADDW	V11, V15, V11
+	VXORV	V4, V8, V4
+	VXORV	V5, V9, V5
+	VXORV	V6, V10, V6
+	VXORV	V7, V11, V7
+	VROTRW	$20, V4, V4
+	VROTRW	$20, V5, V5
+	VROTRW	$20, V6, V6
+	VROTRW	$20, V7, V7
+
+	// V0..V3 += V4..V7
+	// V12..V15 <<<= ((V12..V15 XOR V0..V3), 8)
+	VADDW	V0, V4, V0
+	VADDW	V1, V5, V1
+	VADDW	V2, V6, V2
+	VADDW	V3, V7, V3
+	VXORV	V12, V0, V12
+	VXORV	V13, V1, V13
+	VXORV	V14, V2, V14
+	VXORV	V15, V3, V15
+	VROTRW	$24, V12, V12
+	VROTRW	$24, V13, V13
+	VROTRW	$24, V14, V14
+	VROTRW	$24, V15, V15
+
+	// V8..V11 += V12..V15
+	// V4..V7 <<<= ((V4..V7 XOR V8..V11), 7)
+	VADDW	V12, V8, V8
+	VADDW	V13, V9, V9
+	VADDW	V14, V10, V10
+	VADDW	V15, V11, V11
+	VXORV	V4, V8, V4
+	VXORV	V5, V9, V5
+	VXORV	V6, V10, V6
+	VXORV	V7, V11, V7
+	VROTRW	$25, V4, V4
+	VROTRW	$25, V5, V5
+	VROTRW	$25, V6, V6
+	VROTRW	$25, V7, V7
+
+	// V0..V3 += V5..V7, V4
+	// V15,V12-V14 <<<= ((V15,V12-V14 XOR V0..V3), 16)
+	VADDW	V0, V5, V0
+	VADDW	V1, V6, V1
+	VADDW	V2, V7, V2
+	VADDW	V3, V4, V3
+	VXORV	V15, V0, V15
+	VXORV	V12, V1, V12
+	VXORV	V13, V2, V13
+	VXORV	V14, V3, V14
+	VROTRW	$16, V15, V15
+	VROTRW	$16, V12, V12
+	VROTRW	$16, V13, V13
+	VROTRW	$16, V14, V14
+
+	// V10,V11,V8,V9 += V15,V12,V13,V14
+	// V5,V6,V7,V4 <<<= ((V5,V6,V7,V4 XOR V10,V11,V8,V9), 12)
+	VADDW	V10, V15, V10
+	VADDW	V11, V12, V11
+	VADDW	V8, V13, V8
+	VADDW	V9, V14, V9
+	VXORV	V5, V10, V5
+	VXORV	V6, V11, V6
+	VXORV	V7, V8, V7
+	VXORV	V4, V9, V4
+	VROTRW	$20, V5, V5
+	VROTRW	$20, V6, V6
+	VROTRW	$20, V7, V7
+	VROTRW	$20, V4, V4
+
+	// V0..V3 += V5..V7, V4
+	// V15,V12-V14 <<<= ((V15,V12-V14 XOR V0..V3), 8)
+	VADDW	V5, V0, V0
+	VADDW	V6, V1, V1
+	VADDW	V7, V2, V2
+	VADDW	V4, V3, V3
+	VXORV	V15, V0, V15
+	VXORV	V12, V1, V12
+	VXORV	V13, V2, V13
+	VXORV	V14, V3, V14
+	VROTRW	$24, V15, V15
+	VROTRW	$24, V12, V12
+	VROTRW	$24, V13, V13
+	VROTRW	$24, V14, V14
+
+	// V10,V11,V8,V9 += V15,V12,V13,V14
+	// V5,V6,V7,V4 <<<= ((V5,V6,V7,V4 XOR V10,V11,V8,V9), 7)
+	VADDW	V15, V10, V10
+	VADDW	V12, V11, V11
+	VADDW	V13, V8, V8
+	VADDW	V14, V9, V9
+	VXORV	V5, V10, V5
+	VXORV	V6, V11, V6
+	VXORV	V7, V8, V7
+	VXORV	V4, V9, V4
+	VROTRW	$25, V5, V5
+	VROTRW	$25, V6, V6
+	VROTRW	$25, V7, V7
+	VROTRW	$25, V4, V4
+
+	SUBV	$1, R15
+	BNE	R15, R0, chacha
+
+	// load origin contants
+	VMOVQ	(R10), V16.W4
+	VMOVQ	4(R10), V17.W4
+	VMOVQ	8(R10), V18.W4
+	VMOVQ	12(R10), V19.W4
+
+	// load origin keys
+	VMOVQ	(R7), V20.W4
+	VMOVQ	4(R7), V21.W4
+	VMOVQ	8(R7), V22.W4
+	VMOVQ	12(R7), V23.W4
+	VMOVQ	16(R7), V24.W4
+	VMOVQ	20(R7), V25.W4
+	VMOVQ	24(R7), V26.W4
+	VMOVQ	28(R7), V27.W4
+
+	// add back the initial state to generate the key stream
+	VADDW	V30, V12, V12	// update counter in advance to prevent V30 from being overwritten
+	VADDW	V16, V0, V0
+	VADDW	V17, V1, V1
+	VADDW	V18, V2, V2
+	VADDW	V19, V3, V3
+
+	// load origin counter + nonce
+	VMOVQ	(R9), V28.W4
+	VMOVQ	(R8), V29.W4
+	VMOVQ	4(R8), V30.W4
+	VMOVQ	8(R8), V31.W4
+
+	VADDW	V20, V4, V4
+	VADDW	V21, V5, V5
+	VADDW	V22, V6, V6
+	VADDW	V23, V7, V7
+	VADDW	V24, V8, V8
+	VADDW	V25, V9, V9
+	VADDW	V26, V10, V10
+	VADDW	V27, V11, V11
+	VADDW	V28, V12, V12
+	VADDW	V29, V13, V13
+	VADDW	V30, V14, V14
+	VADDW	V31, V15, V15
+
+	// shuffle
+	VILVLW	V0, V1, V16
+	VILVHW	V0, V1, V17
+	VILVLW	V2, V3, V18
+	VILVHW	V2, V3, V19
+	VILVLW	V4, V5 ,V20
+	VILVHW	V4, V5, V21
+	VILVLW	V6, V7, V22
+	VILVHW	V6, V7, V23
+	VILVLW	V8, V9, V24
+	VILVHW	V8, V9, V25
+	VILVLW	V10, V11, V26
+	VILVHW	V10, V11, V27
+	VILVLW	V12, V13, V28
+	VILVHW	V12, V13, V29
+	VILVLW	V14, V15, V30
+	VILVHW	V14, V15, V31
+	VILVLV	V16, V18, V0
+	VILVHV	V16, V18, V4
+	VILVLV	V17, V19, V8
+	VILVHV	V17, V19, V12
+
+	// load src data from R5
+	VMOVQ	0(R5), V16
+	VMOVQ	16(R5), V17
+	VMOVQ	32(R5), V18
+	VMOVQ	48(R5), V19
+
+	VILVLV	V20, V22, V1
+	VILVHV	V20, V22, V5
+	VILVLV	V21, V23, V9
+	VILVHV	V21, V23, V13
+
+	VMOVQ	64(R5), V20
+	VMOVQ	80(R5), V21
+	VMOVQ	96(R5), V22
+	VMOVQ	112(R5), V23
+
+	VILVLV	V24, V26, V2
+	VILVHV	V24, V26, V6
+	VILVLV	V25, V27, V10
+	VILVHV	V25, V27, V14
+
+	VMOVQ	128(R5), V24
+	VMOVQ	144(R5), V25
+	VMOVQ	160(R5), V26
+	VMOVQ	176(R5), V27
+
+	VILVLV	V28, V30, V3
+	VILVHV	V28, V30, V7
+	VILVLV	V29, V31, V11
+	VILVHV	V29, V31, V15
+
+	VMOVQ	192(R5), V28
+	VMOVQ	208(R5), V29
+	VMOVQ	224(R5), V30
+	VMOVQ	240(R5), V31
+
+	VXORV	V0, V16, V16
+	VXORV	V1, V17, V17
+	VXORV	V2, V18, V18
+	VXORV	V3, V19, V19
+
+	VMOVQ	V16, 0(R4)
+	VMOVQ	V17, 16(R4)
+	VMOVQ	V18, 32(R4)
+	VMOVQ	V19, 48(R4)
+
+	VXORV	V4, V20, V20
+	VXORV	V5, V21, V21
+	VXORV	V6, V22, V22
+	VXORV	V7, V23, V23
+
+	VMOVQ	V20, 64(R4)
+	VMOVQ	V21, 80(R4)
+	VMOVQ	V22, 96(R4)
+	VMOVQ	V23, 112(R4)
+
+	VXORV	V8, V24, V24
+	VXORV	V9, V25, V25
+	VXORV	V10, V26, V26
+	VXORV	V11, V27, V27
+
+	VMOVQ	V24, 128(R4)
+	VMOVQ	V25, 144(R4)
+	VMOVQ	V26, 160(R4)
+	VMOVQ	V27, 176(R4)
+
+	VXORV	V12, V28, V28
+	VXORV	V13, V29, V29
+	VXORV	V14, V30, V30
+	VXORV	V15, V31, V31
+
+	VMOVQ	V28, 192(R4)
+	VMOVQ	V29, 208(R4)
+	VMOVQ	V30, 224(R4)
+	VMOVQ	V31, 240(R4)
+
+	ADD	$4, R12, R12
+	MOVW	R12, (R9) // update counter
+
+	ADDV	$256, R4, R4
+	ADDV	$256, R5, R5
+	SUBV    $256, R6, R6
+	BNE     R6, R0, loop
+
+	RET
diff --git a/chacha20/chacha_noasm.go b/chacha20/chacha_noasm.go
index c709b72..3853cc0 100644
--- a/chacha20/chacha_noasm.go
+++ b/chacha20/chacha_noasm.go
@@ -2,7 +2,7 @@
 // Use of this source code is governed by a BSD-style
 // license that can be found in the LICENSE file.
 
-//go:build (!arm64 && !s390x && !ppc64 && !ppc64le) || !gc || purego
+//go:build (!arm64 && !loong64 && !s390x && !ppc64 && !ppc64le) || !gc || purego
 
 package chacha20