Compare
DeepSeek V3 0324 vs DeepSeek-V3.2
Put two to four models side by side on the fields that decide a deployment: context, price, region, license, languages, hosting, and what is verified.
| DeepSeek V3 0324 | DeepSeek-V3.2 | |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Parameters | 684.5B | 685.36B |
| Context window | Not verified: the provider has not documented this claim. | 164k |
| License | mit | mit |
| Modalities | text | text |
| Release date | 24 Mar 2025 | 01 Dec 2025 |
| Languages | Not verified: the provider has not documented this claim. | Strong: en, zh Adequate: de, fr, es, it, pt, ru, ja, ko, ar |
| Open weights | yes | yes |
| Tasks | Reasoning, Coding | Reasoning, Coding, Summarisation, Retrieval (RAG) |
Where two models differ, this table shows the difference and stops. LLM EU publishes no winner badge without a documented score for that task, and a missing figure stays missing instead of becoming a zero.
When to use it
DeepSeek V3 0324
Strengths
- open weights are published on Hugging Face and can be downloaded without a vendor agreement
- mit licence, so commercial self-hosting is permitted without a separate agreement
Limits
- catalogue only — LLM EU does not run this model, and nothing on this card implies an LLM EU endpoint
- no evaluation has been run by LLM EU
- the published configuration records no context window, so the card carries none
- at 684.5B parameters, self-hosting needs a multi-node deployment rather than a single server
Tasks
DeepSeek-V3.2
Strengths
- MIT licence
- sparse attention aimed at lower long-context inference cost
- 163,840-token configuration maximum
Limits
- 685B parameters: full-precision serving needs several nodes
- vendor efficiency claims are not independently reproduced here
- no evaluation has been run by LLM EU