DeepSeek V3 0324
DeepSeek · released Mar 24, 2025 · deepseek-ai/DeepSeek-V3-0324
- Type
- Open weightsMIT License
- Params
- 685B
- Context
- 164K
37B active per word · about 123K words of context
Our take
Written Sep 2, 2026DeepSeek V3 is a large downloadable text model released in March 2025 with a permissive MIT licence. It carries 685 billion parameters with 37 billion active per token, and handles up to 163,840 tokens in a single request. Its coding score sits well above its other measured capabilities, making it a code-biased generalist rather than an all-rounder.
Pick this for open-weights deployment with a genuinely permissive licence, or cost-sensitive production text work where coding dominates. Use it for long-document tasks up to 163,840 tokens, or where provider choice drives price competition. Skip it if you need image, video or audio input, if maths or precise instruction following dominate, or if you need guaranteed throughput — half the tracked offers lack speed data and the rest vary widely.
The case for it
- Coding performance is 32.9 points above its own overall text score, and 60.1 points above its weakest measured area.
- MIT licence allows commercial use, modification and redistribution without copyleft requirements.
- Eight offers from six providers, with the cheapest entry pricing roughly half the most expensive.
- 163,840-token request limit supports substantial long-document work.
The case against it
- Maths is its weakest measured area, 60.1 points below coding and 27.2 points below overall text.
- Throughput data exists for only four of eight offers, and measured speeds span a 65% range.
- Text-only: no image, video or audio input or output.
How good is it?
EverydayGeneral questions and everyday reasoning
Arena Text (overall)99th of 168 · 1396
CodingWriting and fixing code on its own
Arena Coding110th of 168 · 1428
Arena Coding is the only board that has scored it for this.
AgenticPlanning, calling tools, staying on task
Not yet scored on Arena Agent.
WritingDrafting and rewriting prose
Arena Creative Writing76th of 168 · 1390
Arena Creative Writing is the only board that has scored it for this.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done.
Every published score for this model6 scoresEvery figure we hold, from 6 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
GeForce RTX 4090 · 24 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Radeon RX 7900 XT · 20 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Memory use by level
Against a 24 GB card.
What is quantisation? →This model on every device we track71 devicesThe Q4 build most people download, on each device: what the weights come to, how much context the memory leaves, and whether it runs. Smallest device that runs it first.
Check against your own machine → · Where to rent it hosted →
Or rent it from someone else
Prices checked between 2 hours and 21 days ago — each listing carries its own date.
Novita AI, direct
Cheapest of the 2 listings we can compare like for like — at 164K of context, out of 5 in the table below. 2 cheaper rows there are outside that comparison: a different quantisation.
- per 1M tokens
- $0.27 in / $1.12 out
- Context served
- 164K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| DeepInfrafp4Direct | $0.24 / $0.90checked 9 days ago | 164K | not measured | Unknown | Unknown | Unknown |
| SiliconFlowfp8Through OpenRouter | $0.25 / $1.00checked 2 hours ago | 164K147K max reply | 20 tok/s | No | No | Confirmed |
| Novita AIDirect | $0.27 / $1.12checked 21 days ago | 164K | not measured | Unknown | Unknown | Unknown |
| OpenRouterOpenRouter's own listing | $0.29 / $1.14checked 2 hours ago | 164K | not measured | Unknown | Unknown | Unknown |
| GMICloudfp8Through OpenRouter | $0.29 / $1.14checked 2 hours ago | 164K147K max reply | 29 tok/s | No | Yesunknown period | Unknown |
DeepInfra
OpenRouterAcross the 5 listings we hold: 2 say they do not train on prompts, 0 say they do and 3 do not say. 1 appears in the zero-retention registry we check; the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| DeepInfrafp4Direct | |||
| SiliconFlowfp8Through OpenRouter | ✓ | ✓ | ✓ |
| Novita AIDirect | |||
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| GMICloudfp8Through OpenRouter | ✗ | ✓ | ✓ |
Tool calling: 2 of 5 listings say yes, 1 says no, 2 publish no parameter list. JSON output: 3 of 5 listings say yes, 2 publish no parameter list. Strict schema: 3 of 5 listings say yes, 2 publish no parameter list.
Models people weigh against DeepSeek V3 0324
When we formed this view
Recent changes
What moved
first indexed by our pipelineEach date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- 2 of 5 listings publish no parameter list, so what their API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 3 of 5 listings do not say whether they train on prompts.
- We hold no batch or off-peak rate for any of its listings.
- We hold a decode speed for it, but no prompt-processing (prefill) figure, so how long the input side of a job takes is unknown to us.
Licence and identifiers
What the licence allowsMIT License, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
MIT License
Fully permissive: do anything with attribution. No patent grant, unlike Apache-2.0.
Identifiers
- Hugging Face
- deepseek-ai/DeepSeek-V3-0324
- Architecture
- Mixture of experts
- Takes in, gives back
- Text in, text out
- Catalogue slug
- deepseek-deepseek-v3-0324