Total runs:
22.8K
Run Growth:
18.1K
Growth Rate:
80.25%
Sponsored by
APIMart
Solstice-AI / DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark
Model's Last Updated: September 11 2026
image-text-to-text
Introduction of DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark
Model Details of DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark
DeepSeek-V4-Flash-Vision-Exp (oQ4e Mixed-Precision)
Official Solstice-AI Apple Silicon Release • oQ4e Universal Dynamic Quantization • Native 1M YaRN Context (1,048,576 Tokens) • Up to 31+ tok/s on Apple Silicon
Original Architecture by DeepSeek-AI • Upstream Checkpoint by Unsloth • oQ4e Mixed-Precision Optimization by Solstice-AI
-purple){kind=icon}
Model Summary
Solstice-AI/DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e
is the official
oQ4e
(oMLX Universal Dynamic Quantization) mixed-precision release of DeepSeek's flagship multimodal foundation model,
DeepSeek-V4-Flash-Vision-Exp
(284B total parameters, 256 routed MoE experts, ~13B active per token).
Mixed-Precision Quantization Details:
- Quantization Format: oMLX Universal Dynamic Quantization (oQ4e).
- Base Precision: 4-bit affine (group_size=64).
- Target bpw: ~4.6 bpw.
-
Consensus-Critical Layer Protection:
-
lm_head: strictly protected at 8-bit within budget. -
MoE Routers & Gate Projections (
mlp.gate,gate,shared_expert_gate): protected at full precision / 8-bit to preserve expert routing fidelity. - 32-Layer Vision Tower ViT & Multimodal Aligner: kept in untouched full BF16 .
-
Attention Sinks & Hyper-Connection Tables (
hc_mult: 4): kept in full BF16/FP32 .
-
-
Sensitivity-Driven Dynamic Boosts:
- Sensitive attention projections dynamically boosted up to 8-bit based on Hessian error gradients.
- Routed MoE experts remain at base 4-bit , achieving maximum generation throughput on Apple Silicon unified memory.
-
Native 1M Context Window:
Exact untouched 1,048,576 tokens (
rope_scaling: factor 16, typeyarn, original context 65,536 tokens,rope_theta: 10000).
Official DeepSeek-V4-Flash Benchmark Scoreboard
| Benchmark Suite | Discipline | DeepSeek-V4-Flash MLX | Claude 3.5 Sonnet | GPT-4o |
|---|---|---|---|---|
| Terminal-Bench 2.1 | Agentic Terminal / CLI Execution | 83.9% | 63.5% | 58.7% |
| SWE-bench Verified | Real-World Software Engineering | 65.8% | 61.2% | 48.9% |
| LiveCodeBench v6 | Competitive Algorithmic Coding | 84.2% | 78.4% | 72.8% |
| MATH-500 | High-School / Olympiad Math | 94.6% | 89.2% | 91.4% |
| AIME 2025 | American Invitational Mathematics Exam | 78.2% | 72.5% | 63.8% |
| MMMU (Multimodal) | Multi-Discipline Visual Understanding | 71.4% | 70.4% | 69.1% |
| DocVQA / ChartQA | Complex Document & Graph Reasoning | 92.3% | 91.8% | 89.5% |
Quickstart on Apple Silicon
1. Requirements
- Apple Silicon Mac (M2 Ultra, M3 Ultra, M4 Max/Ultra recommended for full 1M context serving)
- macOS 15.0 (Sequoia) or later
-
Python 3.10+ and
mlx/mlx-lm
pip install mlx mlx-lm huggingface_hub
2. Loading with
mlx-lm
from mlx_lm import load, generate
model, tokenizer = load("Solstice-AI/DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e")
response = generate(model, tokenizer, prompt="Explain Multi-Head Latent Attention in DeepSeek-V4.", max_tokens=1024, verbose=True)
print(response)
3. Native Context Configuration
{
"max_position_embeddings": 1048576,
"rope_scaling": {
"beta_fast": 32,
"beta_slow": 1,
"factor": 16,
"original_max_position_embeddings": 65536,
"type": "yarn"
},
"rope_theta": 10000
}
License & Attribution
- Base Architecture: DeepSeek-AI ( DeepSeek-V4 License )
- Packaging & Optimization: Solstice-AI
Runs of Solstice-AI DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark on huggingface.co
376
Total runs
0
24-hour runs
53
3-day runs
376
7-day runs
376
30-day runs
More Information About DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark huggingface.co Model
Url of DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark
DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark huggingface.co Url
Provider of DeepSeek-V4-Flash-Vision-Exp-mlx-oQ4e-DSpark huggingface.co
Other API from Solstice-AI
Updated:October 15 2026
Total runs:
17.5K
Run Growth:
7.3K
Growth Rate:
41.91%
Total runs:
16.1K
Run Growth:
-646
Growth Rate:
-4.01%
Total runs:
9.8K
Run Growth:
1.6K
Growth Rate:
16.60%
Total runs:
8.7K
Run Growth:
8.7K
Growth Rate:
100.00%
Total runs:
7.8K
Run Growth:
3.1K
Growth Rate:
40.00%
Total runs:
7.6K
Run Growth:
2.0K
Growth Rate:
26.57%
Total runs:
6.0K
Run Growth:
5.3K
Growth Rate:
87.65%
Total runs:
4.3K
Run Growth:
3.6K
Growth Rate:
83.25%
Total runs:
3.2K
Run Growth:
1.5K
Growth Rate:
46.94%
Total runs:
3.1K
Run Growth:
-8.1K
Growth Rate:
-255.42%
Total runs:
3.1K
Run Growth:
1.2K
Growth Rate:
38.73%
Total runs:
2.8K
Run Growth:
2.8K
Growth Rate:
100.00%
Total runs:
2.4K
Run Growth:
2.4K
Growth Rate:
100.00%
Total runs:
2.3K
Run Growth:
2.3K
Growth Rate:
97.27%
Total runs:
2.0K
Run Growth:
874
Growth Rate:
42.70%
Total runs:
2.0K
Run Growth:
-2.4K
Growth Rate:
-117.98%
Total runs:
1.9K
Run Growth:
-9.5K
Growth Rate:
-488.58%
Total runs:
1.7K
Run Growth:
1.4K
Growth Rate:
83.28%
Total runs:
1.6K
Run Growth:
-7.8K
Growth Rate:
-457.85%
Total runs:
1.6K
Run Growth:
1.1K
Growth Rate:
65.47%
Total runs:
1.5K
Run Growth:
684
Growth Rate:
45.84%
Total runs:
1.4K
Run Growth:
1.1K
Growth Rate:
83.01%
Total runs:
1.4K
Run Growth:
1.1K
Growth Rate:
82.08%
Total runs:
1.3K
Run Growth:
1.3K
Growth Rate:
100.00%
Total runs:
1.3K
Run Growth:
963
Growth Rate:
75.23%
Total runs:
1.1K
Run Growth:
1.1K
Growth Rate:
100.00%
Total runs:
1.1K
Run Growth:
-1.0K
Growth Rate:
-91.95%
Total runs:
1.0K
Run Growth:
915
Growth Rate:
100.00%
Total runs:
1.0K
Run Growth:
357
Growth Rate:
35.28%
Total runs:
983
Run Growth:
983
Growth Rate:
100.00%
Total runs:
982
Run Growth:
580
Growth Rate:
59.06%
Total runs:
941
Run Growth:
407
Growth Rate:
43.25%
Total runs:
938
Run Growth:
631
Growth Rate:
67.27%
Total runs:
879
Run Growth:
-706
Growth Rate:
-79.86%
Total runs:
811
Run Growth:
536
Growth Rate:
66.09%
Total runs:
752
Run Growth:
752
Growth Rate:
100.00%
Total runs:
744
Run Growth:
657
Growth Rate:
88.31%
Total runs:
731
Run Growth:
105
Growth Rate:
14.36%
Total runs:
724
Run Growth:
601
Growth Rate:
83.01%
Total runs:
712
Run Growth:
327
Growth Rate:
45.93%
Total runs:
645
Run Growth:
-46
Growth Rate:
-6.66%
Total runs:
637
Run Growth:
-800
Growth Rate:
-122.70%
Total runs:
611
Run Growth:
13
Growth Rate:
2.09%
Total runs:
569
Run Growth:
0
Growth Rate:
0.00%
Total runs:
557
Run Growth:
520
Growth Rate:
93.36%
Total runs:
531
Run Growth:
175
Growth Rate:
32.96%
Total runs:
518
Run Growth:
210
Growth Rate:
45.95%
Total runs:
487
Run Growth:
487
Growth Rate:
100.00%
Total runs:
479
Run Growth:
237
Growth Rate:
49.48%
Total runs:
444
Run Growth:
233
Growth Rate:
52.48%
Total runs:
426
Run Growth:
219
Growth Rate:
51.41%
Total runs:
423
Run Growth:
357
Growth Rate:
84.40%
Total runs:
411
Run Growth:
411
Growth Rate:
100.00%
Total runs:
380
Run Growth:
380
Growth Rate:
100.00%
Total runs:
317
Run Growth:
221
Growth Rate:
69.72%
Total runs:
309
Run Growth:
309
Growth Rate:
100.00%
Total runs:
306
Run Growth:
306
Growth Rate:
100.00%
Total runs:
305
Run Growth:
259
Growth Rate:
84.92%
Total runs:
304
Run Growth:
-149
Growth Rate:
-48.69%
Total runs:
293
Run Growth:
293
Growth Rate:
100.00%
Total runs:
286
Run Growth:
286
Growth Rate:
100.00%
Total runs:
261
Run Growth:
218
Growth Rate:
83.52%
Total runs:
235
Run Growth:
235
Growth Rate:
100.00%
Total runs:
220
Run Growth:
220
Growth Rate:
100.00%
Total runs:
187
Run Growth:
187
Growth Rate:
100.00%
Total runs:
162
Run Growth:
162
Growth Rate:
100.00%
Total runs:
110
Run Growth:
38
Growth Rate:
34.55%
Total runs:
84
Run Growth:
84
Growth Rate:
100.00%
Total runs:
57
Run Growth:
-6
Growth Rate:
-10.53%
Total runs:
45
Run Growth:
0
Growth Rate:
0.00%
Total runs:
40
Run Growth:
0
Growth Rate:
0.00%
Total runs:
1
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%
Total runs:
0
Run Growth:
0
Growth Rate:
0.00%