Overview

Identity

Identity block not returned by the API for this entity.

Openness

Openness not classified yet — no sourced evidence to place this model in the ontology.

Key facts

Release date: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
Model card: huggingface.co/deepseek-ai/DeepSeek-V4-Flash-DSpark
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium

Architecture

Architecture: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
Model type: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
Parameters: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
Weights dtype: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
File size: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
Library name: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
Pipeline tag: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium
Hugging Face repo: Source:Hugging Face Hub (public pages, model cards, papers)T2observed 12 h agomedium

Hardware fit37

Estimated

15 of 37 device × quantization combinations fit.

Run locally: your machine →

Estimated hardware fit
Hardware	Quantization	Device memory	Est. memory	Fits
Apple M3 Ultra	4bit	—	95.5 GB est.	Yes
Mac Studio (Apple M5 Ultra)	4bit	—	95.5 GB est.	Yes
Apple M2 Ultra	4bit	—	95.5 GB est.	Yes
Apple M1 Ultra	4bit	—	95.5 GB est.	Yes
Apple M3 Max	4bit	—	95.5 GB est.	Yes
Apple M4 Max	4bit	—	95.5 GB est.	Yes
Mac Studio (Apple M5 Max)	4bit	—	95.5 GB est.	Yes
MacBook Pro (Apple M5 Max)	4bit	—	95.5 GB est.	Yes
NVIDIA DGX Spark	4bit	128 GB	95.5 GB est.	Yes
NVIDIA H200	4bit	141 GB	95.5 GB est.	Yes
NVIDIA H200 NVL	4bit	141 GB	95.5 GB est.	Yes
NVIDIA B200	4bit	180 GB	95.5 GB est.	Yes
AMD Instinct MI300X	4bit	192 GB	95.5 GB est.	Yes
AMD Instinct MI325X	4bit	256 GB	95.5 GB est.	Yes
NVIDIA DGX B200	4bit	1,440 GB	95.5 GB est.	Yes
Apple M2 Max	4bit	—	95.5 GB est.	No
Apple M1 Max	4bit	—	95.5 GB est.	No
Apple M4 Pro	4bit	—	95.5 GB est.	No
Mac mini (Apple M5 Pro)	4bit	—	95.5 GB est.	No
MacBook Pro (Apple M5 Pro)	4bit	—	95.5 GB est.	No
Apple M3 Pro	4bit	—	95.5 GB est.	No
Apple M1 Pro	4bit	—	95.5 GB est.	No
Apple M2 Pro	4bit	—	95.5 GB est.	No
Apple M4	4bit	—	95.5 GB est.	No
iMac (Apple M4)	4bit	—	95.5 GB est.	No
Mac mini (Apple M6)	4bit	—	95.5 GB est.	No
MacBook Air (Apple M5)	4bit	—	95.5 GB est.	No
MacBook Pro (Apple M5)	4bit	—	95.5 GB est.	No
Apple M2	4bit	—	95.5 GB est.	No
Apple M3	4bit	—	95.5 GB est.	No
Apple M1	4bit	—	95.5 GB est.	No
NVIDIA GeForce RTX 3090	4bit	24 GB	95.5 GB est.	No
NVIDIA GeForce RTX 4090	4bit	24 GB	95.5 GB est.	No
NVIDIA GeForce RTX 5090	4bit	32 GB	95.5 GB est.	No
NVIDIA A100 80GB	4bit	80 GB	95.5 GB est.	No
NVIDIA H100 SXM	4bit	80 GB	95.5 GB est.	No
NVIDIA H100 NVL	4bit	94 GB	95.5 GB est.	No

Assumptions (6)

Estimated, not measured: weights = parameters × bytes/param × 1.15 runtime overhead.
bytes/param: 4bit = 0.5, 8bit = 1.0, fp16 = 2.0 (uniform quantization, no per-layer exceptions).
KV cache approximated at 0.5 GB per 8 192 tokens of context, independent of architecture (GQA/MLA models need less).
A model 'fits' when the estimate is at most the device memory minus 2 GB reserved for the OS and framework.
Mixture-of-experts models are estimated on total parameters (all experts must be resident); active parameters are ignored.
Device memory uses the largest configuration when several are listed (e.g. Apple silicon tiers).

Papers1

arXiv:2606.19348Active
35

Timeline1

Full timeline →

11 Sept 2026
New modelModelDeepSeek-V4-Flash-DSparkDeepSeek
New model: DeepSeek-V4-Flash-DSpark (DeepSeek)
huggingface
huggingface.co

Change history20

Temporal, append-only claims: a new observation closes the previous claim instead of overwriting it. Rewind the record with the as-of picker.

Viewing AI Atlas as of 12 Sept 2026 — attributes exactly as the atlas knew them on that day; later corrections are not shown.

Back to today →

Attributes as of 12 Sept 2026 24 claims in force

Release date: 27 Jun 2026
Openness: open-weights
License: MIT
Architecture: DeepseekV4ForCausalLM
Parameters: 165.3B
Quantization format: fp8
File size: 166.9 GB
Hugging Face repo: deepseek-ai/DeepSeek-V4-Flash-DSpark
Pipeline tag: text-generation
Model card: huggingface.co/deepseek-ai/DeepSeek-V4-Flash-DSpark
Downloads: 729,000
Likes: 271
Commercial use allowed: Yes
Derivatives allowed: Yes
Gated: No
Is quantized: Yes
Last modified: 2026-07-04T03:15:12+00:00
Library name: transformers
Downloads all time: 1,521,636
Model type: deepseek_v4
Redistribution allowed: Yes
Tags: transformers, safetensors, deepseek_v4, text-generation, 8-bit, fp8
Weights available: Yes
Weights dtype: BF16, F32, F8_E4M3, F8_E8M0, I64, I8

20 claims · 20 properties

Release daterelease_date1

Claim history for Release date
Value	Valid from → to	Status	Source	Confidence	Extractor
27 Jun 2026	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Opennessopenness1

Claim history for Openness
Value	Valid from → to	Status	Source	Confidence	Extractor
open-weights	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Licenselicense1

Claim history for License
Value	Valid from → to	Status	Source	Confidence	Extractor
mit	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Architecturearchitecture1

Claim history for Architecture
Value	Valid from → to	Status	Source	Confidence	Extractor
DeepseekV4ForCausalLM	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Parametersparameter_count1

Claim history for Parameters
Value	Valid from → to	Status	Source	Confidence	Extractor
165.3B	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Quantization formatquant_format1

Claim history for Quantization format
Value	Valid from → to	Status	Source	Confidence	Extractor
fp8	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

File sizefile_size_gb1

Claim history for File size
Value	Valid from → to	Status	Source	Confidence	Extractor
166.9 GB	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Hugging Face repohf_repo1

Claim history for Hugging Face repo
Value	Valid from → to	Status	Source	Confidence	Extractor
deepseek-ai/DeepSeek-V4-Flash-DSpark	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Pipeline tagpipeline_tag1

Claim history for Pipeline tag
Value	Valid from → to	Status	Source	Confidence	Extractor
text-generation	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Model cardmodel_card_url1

Claim history for Model card
Value	Valid from → to	Status	Source	Confidence	Extractor
https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-DSpark	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Downloadsmetric.downloads1

Claim history for Downloads
Value	Valid from → to	Status	Source	Confidence	Extractor
729,000	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Likesmetric.likes1

Claim history for Likes
Value	Valid from → to	Status	Source	Confidence	Extractor
271	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Gatedgated1

Claim history for Gated
Value	Valid from → to	Status	Source	Confidence	Extractor
No	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Is quantizedis_quantized1

Claim history for Is quantized
Value	Valid from → to	Status	Source	Confidence	Extractor
Yes	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Last modifiedlast_modified1

Claim history for Last modified
Value	Valid from → to	Status	Source	Confidence	Extractor
2026-07-04T03:15:12+00:00	11 Sept 2026 21:10 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Library namelibrary_name1

Claim history for Library name
Value	Valid from → to	Status	Source	Confidence	Extractor
transformers	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Downloads all timemetric.downloads_all_time1

Claim history for Downloads all time
Value	Valid from → to	Status	Source	Confidence	Extractor
1,521,636	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Model typemodel_type1

Claim history for Model type
Value	Valid from → to	Status	Source	Confidence	Extractor
deepseek_v4	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Tagstags1

Claim history for Tags
Value	Valid from → to	Status	Source	Confidence	Extractor
transformers, safetensors, deepseek_v4, text-generation, 8-bit, fp8	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Weights dtypeweights_dtype1

Claim history for Weights dtype
Value	Valid from → to	Status	Source	Confidence	Extractor
BF16, F32, F8_E4M3, F8_E8M0, I64, I8	11 Sept 2026 21:29 UTC → current	current	Hugging Face Hub (public pages, model cards, papers)T2	medium	deterministic

Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →

Provenance

Attributed facts

20

Source tiers

T220

Freshest observation

6 h ago

Conflicts

None

Source documents 3

Source documents
Source	Document	Type	Tier	Last observed	Snapshots
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/deepseek-ai/DeepSeek-V4-Flash-DSpark/raw/main/README.md	model_card	T2· Quality secondary	5 h ago	1
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/deepseek-ai/DeepSeek-V4-Flash-DSpark	model_page	T2· Quality secondary	6 h ago	5
Hugging Face Hub (public pages, model cards, papers)	huggingface.co/models?author=deepseek-ai&p=0&sort=downloads	listing	T2· Quality secondary	6 h ago	6

Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.

Data quality (53/100) measures how well AI Atlas knows this entity — completeness, primary-source ratio, freshness, conflicts — never how good the model is. Methodology →