Compare
DeepSeek-V4-Flash (0731) vs DeepSeek V4 Pro
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-V4-Flash (0731) | DeepSeek V4 Pro | |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Parameters | 304.18B | 1598.8B |
| Context window | 1.0M | Not verified: the provider has not documented this claim. |
| License | mit | mit |
| Modalities | text | text |
| Release date | 31 Jul 2026 | 22 Apr 2026 |
| Languages | Strong: en, zh Adequate: de, fr, es, it, pt, ru, ja, ko, ar | Not verified: the provider has not documented this claim. |
| Open weights | yes | yes |
| Tasks | Reasoning, Coding, Summarisation, Retrieval (RAG) | Reasoning, Coding |
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-V4-Flash (0731)
Strengths
- MIT licence with no user threshold or field-of-use restriction
- 1M-token context window
- three reasoning-effort levels for latency control
- vendor reports it beats its own larger preview model on published benchmarks
Limits
- 304B parameters require a multi-GPU node even at reduced precision
- all benchmark claims are self-reported by DeepSeek
- no evaluation has been run by LLM EU
Tasks
DeepSeek V4 Pro
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 1598.8B parameters, self-hosting needs a multi-node deployment rather than a single server