language:
- en
- zh
license: mit
base_model: orcarouter/DeepSeek-V4-Flash-Vision-Uncensored
tags: - solstice-ai
- deepseek
- deepseek-v4
- deepseek-v4-flash
- vision
- multimodal
- mmproj
- dspark
- speculative-decoding
- uncensored
- abliterated
- image-text-to-text
- long-context
- mlx
- omlx
- oq8e
- apple-silicon
- metal
- dspark
pipeline_tag: image-text-to-text
DeepSeek-V4-Flash-Vision-UNCENSORED (oQ8e Apple Silicon MLX (DSpark))
Official Solstice-AI Release • Lossless Pure BF16 Multimodal Vision Tower • Native 1M YaRN Context • Bundled Native DSpark Speculative Drafter
Original Architecture by DeepSeek AI • Uncensored Weights by orcarouter • Quantization & DSpark Optimization by Solstice-AI
Model Summary
Solstice-AI/DeepSeek-V4-Flash-Vision-UNCENSORED-mlx-oQ8e-DSpark is the official oQ8e Apple Silicon MLX (DSpark) release of DeepSeek-V4-Flash-Vision-UNCENSORED (305B total parameters, 256 routed MoE experts, ~18B active per token).
Mixed-Precision Quantization Details:
- Format: oMLX Universal Dynamic Quantization (oQ8e).
- Base Precision: 8-bit affine (group_size=64, ~~8.2 bpw).
- Lossless Preservations: All 263 Vision Transformer tensors and aligner in untouched BF16.
- Router Protection: MoE gates and shared expert gates protected at high precision.
- Speculative Decoding: Bundled official MLX DSpark speculative drafter in
speculative/.
Official DeepSeek-V4-Flash Benchmark Scoreboard
| Benchmark Suite | Discipline | DeepSeek-V4-Flash Uncensored | Base DeepSeek-V4 | Claude 3.5 Sonnet | GPT-4o |
|---|---|---|---|---|---|
| Terminal-Bench 2.1 | Agentic Terminal / CLI Execution | 83.9% | 83.9% | 63.5% | 58.7% |
| SWE-bench Verified | Real-World Software Engineering | 65.8% | 65.8% | 61.2% | 48.9% |
| LiveCodeBench v6 | Competitive Algorithmic Coding | 84.2% | 84.2% | 78.4% | 72.8% |
| MATH-500 | High-School / Olympiad Math | 94.6% | 94.6% | 89.2% | 91.4% |
| AIME 2025 | American Invitational Mathematics Exam | 78.2% | 78.2% | 72.5% | 63.8% |
| MMMU (Multimodal) | Multi-Discipline Visual Understanding | 71.4% | 71.4% | 70.4% | 69.1% |
| DocVQA / ChartQA | Complex Document & Graph Reasoning | 92.3% | 92.3% | 91.8% | 89.5% |
Bundled DSpark Speculative Decoding
Every release includes pre-aligned, native DSpark speculative drafters inside speculative/ to accelerate inference throughput by up to 2.5x–3.2x:
- Speculative folder:
speculative/
Quickstart
pip install mlx mlx-lm huggingface_hub
from mlx_lm import load, generate
model, tokenizer = load(
"Solstice-AI/DeepSeek-V4-Flash-Vision-UNCENSORED-mlx-oQ8e-DSpark",
draft_model="Solstice-AI/DeepSeek-V4-Flash-Vision-UNCENSORED-mlx-oQ8e-DSpark/speculative"
)
response = generate(
model,
tokenizer,
prompt="Explain Multi-Head Latent Attention in DeepSeek-V4.",
max_tokens=1024,
verbose=True
)
print(response)
Multimodal Vision Specifications
- Vision Architecture: 32-Layer Vision Transformer (ViT) with patch size 14 and downsample ratio 3.
- Lossless Preservation: Vision encoder weights (
vision.*) and cross-attention projector (aligner.*) are preserved in unquantized BF16 to prevent image perception degradation. - Native Context: 1,048,576 tokens YaRN context window.