Models / DeepSeek V3 0324

DeepSeek V3 0324

MoE workstation
HuggingFace → HuggingBay → DeepSeek → Find for my hardware

671B total, 37B active per token. Q4_K_M requires ~360GB RAM/VRAM. Use Q2_K (~180GB) for DGX Spark.

AI-generated content marks

This model embeds text watermarks in generated text and adds C2PA provenance metadata to supported files such as .png, .jpg, and .svg. Marks can be lost through editing, screenshots, or format conversion, so their absence does not prove a file is human-made.

Provider transparency docs →

coding reasoning
Parameters
671.0B
Context
128k
License
mit
Developer
DeepSeek
Origin
🇨🇳 China
Released
Mar 2025

Scores

Coding
92
Reasoning
88
General
87

Score per dollar

363 pts per $/M input

general_score (87) divided by cheapest input price ($0.24/M). Higher is better value. See live pricing.

Related models

DeepSeek V4.1 Flash
552.0B 16.0B active workstation 265GB min RAM $0.02/M in Sep 10, 2026
DeepSeek V4 Flash 0731
284.0B 13.0B active workstation 168GB min RAM $0.01/M in Jul 31, 2026
DeepSeek V4 Flash superseded
284.0B 13.0B active workstation no local build $0.01/M in Apr 24, 2026
DeepSeek V4 Pro
1600.0B 49.0B active premier no local build $0.25/M in Apr 24, 2026
DeepSeek V3.1 Terminus
685.0B 37.0B active workstation no local build $0.27/M in Sep 29, 2025
DeepSeek V3.2 Exp
685.0B 37.0B active workstation no local build $0.13/M in Sep 29, 2025

Save your hardware and every model page answers the real question: will it run on your machine, and how fast?

Join free - save your rig →

Run it locally

Per-quant memory needs and a static "can you run it?" reference - no rig entry required

Q2_K
188.0GB 188.0GB min 210.0GB rec
Smallest footprint, noticeable quality loss
Q4_K_M
368.0GB 368.0GB min 400.0GB rec
Balanced - the usual local sweet spot

The reference hardware

Schematic of the NVIDIA Jetson Orin NX 16GB reference rig - 16GB unified memory, 102 GB/s aggregate bandwidth
Schematic of the Jetson AGX Orin 64GB reference rig - 64GB unified memory, 204 GB/s aggregate bandwidth
Schematic of the Single GTX 1080 Ti (11GB) reference rig - 11GB VRAM, 484 GB/s aggregate bandwidth
Schematic of the Single RTX 4090 (24GB) reference rig - 24GB VRAM, 1008 GB/s aggregate bandwidth
Schematic of the Single RTX 5090 (32GB) reference rig - 32GB VRAM, 1792 GB/s aggregate bandwidth
Schematic of the RTX PRO 6000 Blackwell (96GB) reference rig - 96GB VRAM, 1792 GB/s aggregate bandwidth
Schematic of the 2x RTX 3090 (48GB) reference rig - 48GB VRAM, 1872 GB/s aggregate bandwidth
Schematic of the 2x RTX 5090 (64GB) reference rig - 64GB VRAM, 3584 GB/s aggregate bandwidth
Schematic of the 4x RTX 4090 (96GB) reference rig - 96GB VRAM, 4032 GB/s aggregate bandwidth
Schematic of the 4x RTX 5090 (128GB) reference rig - 128GB VRAM, 7168 GB/s aggregate bandwidth
Schematic of the 8x RTX 3090 rack (192GB) reference rig - 192GB VRAM, 7489 GB/s aggregate bandwidth
Schematic of the AMD Instinct MI300X (192GB) reference rig - 192GB VRAM, 5324 GB/s aggregate bandwidth
Schematic of the MacBook Pro M5 Max 128GB reference rig - 128GB unified memory, 614 GB/s aggregate bandwidth
Schematic of the Ryzen AI Max+ 395 128GB reference rig - 128GB unified memory, 256 GB/s aggregate bandwidth
Schematic of the Mac Studio M4 Ultra 192GB reference rig - 192GB unified memory, 1092 GB/s aggregate bandwidth
Schematic of the DGX Spark 128GB unified reference rig - 128GB unified memory, 273 GB/s aggregate bandwidth
Schematic of the 4x H100 80GB (320GB) reference rig - 320GB VRAM, 13400 GB/s aggregate bandwidth
Schematic of the NVIDIA DGX Station 748GB reference rig - 748GB unified memory, 8000 GB/s aggregate bandwidth
Schematic of the Mac Studio M4 Ultra 512GB reference rig - 512GB unified memory, 1092 GB/s aggregate bandwidth
Schematic of the Dual EPYC 9004 + 768GB DDR5-4800 reference rig - 768GB unified memory, 460 GB/s aggregate bandwidth
Schematic of the Epyc + 512GB DDR4-3200 + 2x RTX 3090 reference rig - 560GB unified memory, 204 GB/s aggregate bandwidth
Schematic of the Epyc + 512GB DDR4-2400 + 2x RTX 3090 reference rig - 560GB unified memory, 153 GB/s aggregate bandwidth

22 reference configs, drawn in-house. Scroll for more.

Can you run it? - reference rigs

Rig Q2_K Q4_K_M
NVIDIA Jetson Orin NX 16GB no -> cloud no -> cloud
Jetson AGX Orin 64GB no -> cloud no -> cloud
Single GTX 1080 Ti (11GB) no -> cloud no -> cloud
Single RTX 4090 (24GB) no -> cloud no -> cloud
Single RTX 5090 (32GB) no -> cloud no -> cloud
RTX PRO 6000 Blackwell (96GB) no -> cloud no -> cloud
2x RTX 3090 (48GB) no -> cloud no -> cloud
2x RTX 5090 (64GB) no -> cloud no -> cloud
4x RTX 4090 (96GB) no -> cloud no -> cloud
4x RTX 5090 (128GB) offload no -> cloud
8x RTX 3090 rack (192GB) tight no -> cloud
AMD Instinct MI300X (192GB) tight no -> cloud
MacBook Pro M5 Max 128GB no -> cloud no -> cloud
Ryzen AI Max+ 395 128GB no -> cloud no -> cloud
Mac Studio M4 Ultra 192GB tight no -> cloud
DGX Spark 128GB unified no -> cloud no -> cloud
4x H100 80GB (320GB) fast 709.8t/s offload
NVIDIA DGX Station 748GB fast 423.8t/s fast 216.7t/s
Mac Studio M4 Ultra 512GB fast 63.1t/s fast 32.3t/s
Dual EPYC 9004 + 768GB DDR5-4800 fast 24.4t/s ok 12.5t/s
Epyc + 512GB DDR4-3200 + 2x RTX 3090 ok 10.9t/s slow 5.6t/s
Epyc + 512GB DDR4-2400 + 2x RTX 3090 ok 8.1t/s slow 4.2t/s

Fit tiers use the same will-it-run logic as the rig finder. For comfortable fits, the badge reflects decode speed: fast >=20 t/s, ok 8-20 t/s, slow <8 t/s. t/s is a bandwidth estimate, not a measured benchmark.

Download options

Q2_K official -20% vs fp16
188.0GB dl 188.0GB min 210.0GB rec
REC RAM vs largest quant
Q4_K_M official -5% vs fp16
368.0GB dl 368.0GB min 400.0GB rec
REC RAM vs largest quant
368GB weights + 32GB overhead + KV cache

Or run it in the cloud

Live per-provider pricing, throughput and uptime - refreshed about 9 hours ago via OpenRouter. Click a column to sort.

Provider Type Input $/M Output $/M Cache $/M Tok/s Latency Uptime Value
GMICloud
API 0.29 1.14 0.110 - - 99.76% best uptime
SiliconFlow
API 0.25 1.00 0.110 - - 99.19%
DeepInfra avoid
API 0.24 0.90 0.135 - - 88.41% cheapest

Default order: throughput among 95%+ uptime providers, then latency; subscriptions last. Sort by any column. Subscription rows show $/mo in the Value column - per-token columns are "-". Affiliate links are marked sponsored / nofollow. Confirm current pricing on the provider's site before committing.

Detailed API pricing page + JSON endpoint →

See who runs DeepSeek in production →

PRICE HISTORY

Inference cost over time

Data accumulates from the first daily sync - longer ranges populate over time. Prices come from OpenRouter snapshots, not a historical API.

Loading price history...