TensorFold/Qwen3.8-Flash-Next-MLX-oQ4-MTP
Image-Text-to-Text • 180B • Updated • 4.68k • 16
Ivan Fioravanti PRO
ivanfioravanti
AI & ML interests
None yet
Recent Activity
liked a model about 7 hours ago
TensorFold/Qwen3.8-Flash-Next-MLX-oQ4-MTP liked a model about 20 hours ago
larryvrh/MiniMax-H3-Turbo-Lora liked a model 13 days ago
FastVideo/FastVideo-FastH3-ComfyOrganizations
liked a
model about 7 hours ago
liked a
model about 20 hours ago
liked a
model 13 days ago
liked a
model 16 days ago
liked a
model 18 days ago
Confirmed on Mac Studio M4 Max 64GB — 38.7 t/s @ 32K, 38.9 t/s @ 64K, zero swap-ins
❤️ 1
12
#2 opened 26 days ago
by 
pswai
Confirmed on a 64GB M5 Pro: 128K OpenCode Coding-Agent Run
❤️ 1
1
#4 opened 24 days ago
by 
alfredoartiles
Getting tool call error with DS4 - Qwen3.8-Flash-Next-DS4-Q4
👍 1
3
#1 opened 28 days ago
by 
Uzinthromby
Benchmark report: Qwen3.8 Flash Next IQ2 on MacBook Pro M5 Pro 64GB
🔥 1
3
#3 opened 25 days ago
by alfredoartiles
Qwen3.8-Flash-Next DS4 IQ2 — 64 GB Mac report (M1 Max)
🔥 1
5
#1 opened 26 days ago
by 
gzorzi1999
updated a
model 26 days ago
published a
model 26 days ago
liked a
model 27 days ago
updated a
model 27 days ago
published a
model about 1 month ago
liked 2
models about 1 month ago
published an article about 1 month ago
Article
Three Million Models and Counting
ivanfioravanti
• liked a
model about 1 month ago
reacted to nightmedia's post with 👍 about 1 month ago
Post
4959
Qwen3.8-27B metrics
It's hard to track all model cards where I post these, so I figured people would get more value out of seeing these in the open.
The performance is as measured on a M4 MBP 128GB, speed may vary depending on your platform.
These are all instruct metrics, generated by including this line in the jinja template:
Then run the test suite to generate the metrics:
This will generate the file:
This is a JSON containing all gathered metrics; for example the q4-hi:
I use the value of acc_norm for metrics, rounded to 3 decimals.
As I get more quants tested, I will add them here.
A complete test run for a single quant takes 7-9 hours depending on quant size, 10-12 hours for BF16 depending on perplexity: this is why you see on my model cards that I usually post the first three, that only take 2-3 hours :)
-G
It's hard to track all model cards where I post these, so I figured people would get more value out of seeing these in the open.
quant arc arc/e boolq hswag obkqa piqa wino
mxfp8 0.591,0.782,0.896,0.746,0.448,0.801,0.711
q8-hi 0.602,0.779,0.896,0.747,0.446,0.793,0.703
q6-hi 0.602,0.775,0.895,0.748,0.448,0.795,0.710
q4-hi 0.604,0.780,0.898,0.744,0.454,0.795,0.708
mxfp4 0.581,0.771,0.889,0.738,0.442,0.798,0.713
1M
mxfp8 0.590,0.787,0.897,0.744,0.446,0.801,0.709
Quant Perplexity Peak Memory Tokens/sec
mxfp8 6.090 ± 0.054 34.74 GB 138
mxfp4 5.952 ± 0.051 21.30 GB 148The performance is as measured on a M4 MBP 128GB, speed may vary depending on your platform.
These are all instruct metrics, generated by including this line in the jinja template:
{%- set enable_thinking = false %}Then run the test suite to generate the metrics:
mlx_lm.evaluate --model MODEL --tasks winogrande boolq arc_challenge arc_easy hellaswag openbookqa piqaThis will generate the file:
eval_MODEL_0.4.9_winogrande_boolq_arc_challenge_arc_easy_hellaswag_openbookqa_piqaThis is a JSON containing all gathered metrics; for example the q4-hi:
"arc_challenge": {
"alias": "arc_challenge",
"acc,none": 0.5819112627986348,
"acc_stderr,none": 0.014413988396996116,
"acc_norm,none": 0.6040955631399317,
"acc_norm_stderr,none": 0.01429122839353657
},I use the value of acc_norm for metrics, rounded to 3 decimals.
As I get more quants tested, I will add them here.
A complete test run for a single quant takes 7-9 hours depending on quant size, 10-12 hours for BF16 depending on perplexity: this is why you see on my model cards that I usually post the first three, that only take 2-3 hours :)
-G
- 30 replies