DeepSeek-V2.5

DeepSeek · · 122 days after DeepSeek-V2

Verified

2 of 2 published values traced to a source2 primary sources, 2 archived1 not disclosed by the lab

Why this status

Cited to official documentation and the official model card, and every value it asserts was found there. Modalities recorded as text-to-text. No image, audio or video capability appears anywhere in the primary sources, which labs advertise prominently when present. Capabilities evidenced from the primary sources: coding: “5: A New Open-Source Model Combining General and Coding Capabilities | DeepSeek API Docs Skip to main content DeepSeek API Docs English English 中文(中国) DeepSeek Platform Q” | function_calling: “5 Release 2024/09/05 Context Caching is Available 2024/08/02 New API Features 2024/07/25 API Reference API Guides Multi-round Conversation Chat Prefix Completion (Beta) F” | structured_output: “5 Release 2024/09/05 Context Caching is Available 2024/08/02 New API Features 2024/07/25 API Reference API Guides Multi-round Conversation Chat Prefix Completion (Beta) F”.

What these states mean · How records are judged · Has this record changed?

An all-in-one merge of DeepSeek-V2-Chat and DeepSeek-Coder-V2 that combines general conversational and code capabilities.

codingfunction callingstructured outputopen-weights

Details

ReleasedSeptember 5, 2024
stated in huggingface.co

DeepSeek-V2.5 was released on September 5, 2024.

CompanyDeepSeek
FamilyDeepSeek
Lineageafter DeepSeek-V2 · before DeepSeek-V3
EraMultimodal & open weights
TypeLanguage model · LLM
ModalitiesText → Text
CapabilitiesCoding, Function calling, Structured output
WeightsOpen weights
LicenceDeepSeek License Agreement 1.0
Context windowNot disclosed
Parameters236B
stated in huggingface.co

DeepSeek-V2.5 has 236,000,000,000 parameters.

Current information, elsewhere

This record is a historical one: every figure above is traced to a dated primary source. For what is true today — live pricing, current rankings, availability — these are better maintained than anything here could be.

Sources

Record status: verified · confidence 90/100 · 2 of 2 primary

Cited to official documentation and the official model card, and every value it asserts was found there. Modalities recorded as text-to-text. No image, audio or video capability appears anywhere in the primary sources, which labs advertise prominently when present. Capabilities evidenced from the primary sources: coding: “5: A New Open-Source Model Combining General and Coding Capabilities | DeepSeek API Docs Skip to main content DeepSeek API Docs English English 中文(中国) DeepSeek Platform Q” | function_calling: “5 Release 2024/09/05 Context Caching is Available 2024/08/02 New API Features 2024/07/25 API Reference API Guides Multi-round Conversation Chat Prefix Completion (Beta) F” | structured_output: “5 Release 2024/09/05 Context Caching is Available 2024/08/02 New API Features 2024/07/25 API Reference API Guides Multi-round Conversation Chat Prefix Completion (Beta) F”.

Primary means published by the organisation that made the model. Verified records require at least one.

DeepSeek family

  1. DeepSeek-V2 May 6, 2024
  2. DeepSeek-V2.5 September 5, 2024
  3. DeepSeek-V3 December 26, 2024
  4. DeepSeek-R1 January 20, 2025
  5. DeepSeek-V3.1 August 21, 2025
  6. DeepSeek V4 Flash April 24, 2026
  7. DeepSeek-V4.1-Flash September 10, 2026

What changed from DeepSeek-V2

Compared with the previous release in this series. A field appears only when both records state a value — where one does not, it is listed as uncomparable rather than dropped, because a gap in our research is not a finding about the model.

Not comparable: Context window (not disclosed for both releases)

Compare them side by side →

Browse models

Compare DeepSeek-V2.5 with another model →
See DeepSeek-V2.5 on the interactive timeline →