qwen3-0.6b-int8

A Core AI bundle of Qwen/Qwen3-0.6B for the Aether SDK (iOS and macOS 27+).

• Source: Qwen/Qwen3-0.6B at revision c1899de289a04d12100db370d81485cdf75e47ca, licence Apache-2.0. The licence is included as LICENSE.
• Changes from the source: converted from PyTorch to Core AI (.aimodel) by Aether forge (recipe qwen3-0.6b-int8@1). Weights are int8-linear-perchannel (8-bit weights). The tokenizer files are the source's own.

Variants

Variant	Platform	Arch	Compute	Compiled	Assets	Download
macos-any-gpu	macos	any	gpu	no (specialized on first load)	qwen3_0_6b_int8.aimodel 597.5 MB	613.4 MB
ios-any-gpu	ios	any	gpu	no (specialized on first load)	qwen3_0_6b_int8.aimodel 597.5 MB	613.4 MB
ios-h18p-gpu	ios	h18p	gpu	yes	qwen3_0_6b_int8.h18p.aimodelc 597.5 MB	613.4 MB

Verification

Every row is a record in verification/ about exactly these bytes (matched by bundle digest). Reference rows are strict T2 passes of the unquantized export on the same fixture, in verification/reference/.

Variant	Tier	Result	Detail	Device	OS build	Compute	Record
ios-any-gpu	T0	pass		iPhone18,2	24A446	target	6f934189
ios-any-gpu	T2	pass	19/19 strict; profile quantized-8bit; fixture cc6fd71af4ed3af5	iPhone18,2	24A446	target	c8d2ae39
ios-any-gpu	T3	pass	copy-fidelity-v1; 100.0% vs reference 100.0%; 50 items	iPhone18,2	24A446	target	3ec55862
ios-h18p-gpu	T0	pass		iPhone18,2	24A446	target	1858b220
ios-h18p-gpu	T2	pass	19/19 strict; profile quantized-8bit; fixture cc6fd71af4ed3af5	iPhone18,2	24A446	target	42f46a81
ios-h18p-gpu	T3	pass	copy-fidelity-v1; 100.0% vs reference 100.0%; 50 items	iPhone18,2	24A446	target	b313af26
macos-any-gpu	T0	pass		Mac17,6	26A434	target	a268e4d2
macos-any-gpu	T1	pass		Mac17,6	26A434	target	fb8d5aab
macos-any-gpu	T2	pass	18/19 strict; 19/19 strict; profile quantized-8bit; budgeted: think-multiply; fixture cc6fd71af4ed3af5	Mac17,6	26A434	target	cec042a2
macos-any-gpu	T3	pass	copy-fidelity-v1; 100.0% vs reference 100.0%; 50 items	Mac17,6	26A434	target	c965a95c
unquantized reference (not published)	T2	pass	19/19 strict; 19/19 strict; profile strict; fixture cc6fd71af4ed3af5	Mac17,6	26A434	target	e73cf364

The quantized profile also requires: T2 strict on the unquantized reference export (met by the reference row).

Use

aether run qwen3-0.6b-int8 --prompt "Hello"

import Aether

let aether = try Aether()
let chat = try await aether.chat("qwen3-0.6b-int8")
let reply = try await chat.respond(to: "Hello")
print(reply.text)