Open-weight model
Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64
by Rijul Paul rijulpaul/Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64
Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64 is an open-weight model from Rijul Paul. It has 906M parameters. At 16-bit it needs about 2.2 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
Parameters906M
Context—
Weights6.9 GB
License—
AccessOpen weights
Monthly Downloads—
Runs On
What it takes to serve Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64 (906M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 1.8 GB | 2.2 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.9 GB | 1.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.5 GB | 0.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 23, 2026.
Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64 on every accelerator the SAVRN Index prices, at every precision
Model Card
The publisher has not written a card for this model.
Configuration
- Architecture
- Qwen3TTSForConditionalGeneration
- Model type
- qwen3_tts
Identity and Version
- Repository
- rijulpaul/Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64
- Publisher
- Rijul Paul
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- 906M parameters
- Languages
- Not stated by the source
- Revision
- 6d4cfd6f78d7317ce9a87cb1b3606a34008e7650
- First published
- 2026-09-27
- Last updated
- 2026-09-27
Files and Weights
97 files, 6.9 GB in total. The weights are 6 files totalling 6.9 GB in pt, safetensors.
Weights6 files · 6.9 GB
Configuration9 files · 42.1 KB
Tokenizer3 files · 4.5 MB
Other78 files · 21.9 MB
Repository1 file · 5.7 KB
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| adapter/adapter_model.safetensors | Weights | 95.2 MB | f12533b8e6af |
| model.safetensors | Weights | 3.6 GB | b9d076b50b37 |
| speech_tokenizer/model.safetensors | Weights | 682.3 MB | 836b7b357f5e |
| training-checkpoint/pytorch_model/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt | Weights | 285.5 MB | 0dd4cfd1ae1c |
| training-checkpoint/pytorch_model/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt | Weights | 285.5 MB | 17163e692537 |
| training-checkpoint/pytorch_model/mp_rank_00_model_states.pt | Weights | 1.9 GB | b2049060459e |
| adapter/adapter_config.json | Configuration | 1.3 KB | — |
| config.json | Configuration | 4.5 KB | — |
| generation_config.json | Configuration | 245 B | — |
| preprocessor_config.json | Configuration | 127 B | — |
| speech_tokenizer/config.json | Configuration | 2.3 KB | — |
| speech_tokenizer/configuration.json | Configuration | 76 B | — |
| speech_tokenizer/preprocessor_config.json | Configuration | 234 B | — |
| training-checkpoint/train_info.json | Configuration | 70 B | — |
| training-checkpoint/zero_to_fp32.py | Configuration | 33.3 KB | — |
| samples/000/000.wav | Other | 303.4 KB | e681450258c7 |
| samples/000/001.wav | Other | 318.8 KB | 012ffb80ef4d |
| samples/000/002.wav | Other | 215.1 KB | 05bc8a5fd7e1 |
| samples/000/003.wav | Other | 326.4 KB | 73cdd6be727e |
| samples/000/004.wav | Other | 130.6 KB | acae5d39dd0c |
| samples/001/000.wav | Other | 380.2 KB | 57a20fb41b2f |
| samples/001/001.wav | Other | 314.9 KB | b751a68cf0fe |
| samples/001/002.wav | Other | 245.8 KB | 3aead1dd5b80 |
| samples/001/003.wav | Other | 407.1 KB | 8caca94ba1c1 |
| samples/001/004.wav | Other | 134.4 KB | 063ecf327cef |
| samples/002/000.wav | Other | 361.0 KB | f71a700ad3e8 |
| samples/002/001.wav | Other | 368.7 KB | 5c6234c30cb9 |
| samples/002/002.wav | Other | 265.0 KB | 4f89df75ff8e |
| samples/002/003.wav | Other | 349.5 KB | 281e7033d3b8 |
| samples/002/004.wav | Other | 122.9 KB | d0de7e97cea5 |
| samples/003/000.wav | Other | 334.1 KB | 51d448edd0cc |
| samples/003/001.wav | Other | 268.8 KB | 396242b2150e |
| samples/003/002.wav | Other | 230.4 KB | 3ccc788ffa03 |
| samples/003/003.wav | Other | 330.3 KB | 67e4380e5bce |
| samples/003/004.wav | Other | 138.3 KB | 969fbf1f0c88 |
| samples/004/000.wav | Other | 368.7 KB | 1119cb699ad6 |
| samples/004/001.wav | Other | 334.1 KB | 153d4531b0c8 |
| samples/004/002.wav | Other | 249.6 KB | 073ee42dccae |
| samples/004/003.wav | Other | 380.2 KB | 56b24155770b |
| samples/004/004.wav | Other | 146.0 KB | bd378ff36c5c |
| samples/005/000.wav | Other | 338.0 KB | e3f77419dc57 |
| samples/005/001.wav | Other | 303.4 KB | 536319b624fd |
| samples/005/002.wav | Other | 226.6 KB | f07aad29cc1d |
| samples/005/003.wav | Other | 395.6 KB | 8dfe449f1130 |
| samples/005/004.wav | Other | 142.1 KB | f56aec2a6202 |
| samples/006/000.wav | Other | 338.0 KB | faa071cb1204 |
| samples/006/001.wav | Other | 353.3 KB | 42e6727bfe23 |
| samples/006/002.wav | Other | 218.9 KB | ce4e05303782 |
| samples/006/003.wav | Other | 368.7 KB | aaf1bef5f697 |
| samples/006/004.wav | Other | 138.3 KB | 6c514cf4540f |
| samples/007/000.wav | Other | 334.1 KB | a86aabac5526 |
| samples/007/001.wav | Other | 368.7 KB | ff8d8d73edf1 |
| samples/007/002.wav | Other | 203.6 KB | f9cf157ee9a5 |
| samples/007/003.wav | Other | 326.4 KB | a4761c64ddf7 |
| samples/007/004.wav | Other | 134.4 KB | daf49935337a |
| samples/008/000.wav | Other | 311.1 KB | 902135a2ac80 |
| samples/008/001.wav | Other | 322.6 KB | 7cd9c80f838d |
| samples/008/002.wav | Other | 276.5 KB | e998d3f760d0 |
| samples/008/003.wav | Other | 391.7 KB | c6e643b8238a |
| samples/008/004.wav | Other | 165.2 KB | 5050c4ced0ef |
| samples/009/000.wav | Other | 295.7 KB | ed6e6de56a6c |
| samples/009/001.wav | Other | 349.5 KB | cc0552e990cb |
| samples/009/002.wav | Other | 234.3 KB | ce1f43e473d6 |
| samples/009/003.wav | Other | 399.4 KB | 83e928f4a1ee |
| samples/009/004.wav | Other | 169.0 KB | 38a3afb8aa84 |
| samples/010/000.wav | Other | 353.3 KB | 100eed82a7bf |
| samples/010/001.wav | Other | 314.9 KB | ebcdb11745f4 |
| samples/010/002.wav | Other | 341.8 KB | 750a0f15dac3 |
| samples/010/003.wav | Other | 399.4 KB | e8e05913cadb |
| samples/010/004.wav | Other | 142.1 KB | b0306752be08 |
| samples/011/000.wav | Other | 318.8 KB | 8d2c5fa65c0f |
| samples/011/001.wav | Other | 368.7 KB | 8ba103ffd847 |
| samples/011/002.wav | Other | 299.6 KB | f327f9612e55 |
| samples/011/003.wav | Other | 414.8 KB | aef921216891 |
| samples/011/004.wav | Other | 126.8 KB | f430920c2791 |
| samples/012/000.wav | Other | 345.6 KB | 41153ffaac4f |
| samples/012/001.wav | Other | 341.8 KB | a2a810556a32 |
| samples/012/002.wav | Other | 261.2 KB | 646e86261c80 |
| samples/012/003.wav | Other | 410.9 KB | f8492ccf44b0 |
| samples/012/004.wav | Other | 126.8 KB | 8d7711c5dba8 |
| samples/013/000.wav | Other | 326.4 KB | 63f5e3610f23 |
| samples/013/001.wav | Other | 345.6 KB | a8326fbf946c |
| samples/013/002.wav | Other | 261.2 KB | bc7a284f73a5 |
| samples/013/003.wav | Other | 387.9 KB | 0ce2e43e9a22 |
| samples/013/004.wav | Other | 188.2 KB | c85ed0f2f73e |
| samples/014/000.wav | Other | 483.9 KB | 03255e9911ae |
| samples/014/001.wav | Other | 422.4 KB | bbd066d35060 |
| samples/014/002.wav | Other | 272.7 KB | 0cd68dd74c2b |
| samples/014/003.wav | Other | 368.7 KB | 7719da2093d5 |
| samples/014/004.wav | Other | 146.0 KB | 8ffde63c9b40 |
| training-checkpoint/latest | Other | 13 B | — |
| training-checkpoint/random_states_0.pkl | Other | 15.1 KB | e82332f690ef |
| training-checkpoint/random_states_1.pkl | Other | 15.0 KB | 223e4b62448a |
| .gitattributes | Repository | 5.7 KB | — |
| merges.txt | Tokenizer | 1.7 MB | — |
| tokenizer_config.json | Tokenizer | 7.3 KB | — |
| vocab.json | Tokenizer | 2.8 MB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 6.9 GB
Download from Rijul Paul
Released by Rijul Paul through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 6.9 GB |
| 16-bit | 1.8 GB |
| 8-bit | 0.9 GB |
| 4-bit | 0.5 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64
How much GPU memory does Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64 need?
About 2.2 GB at 16-bit and 0.5 GB at 4-bit: the weights (906M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Qwen3-TTS-12Hz-0.6B-Zhongli-LoRa-64 on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.