📋 AI Changelog

DeepSeek

5 active models · 0 deprecated · Official website →

Active Models

DeepSeek V3.2

✅ Active

Latest DeepSeek V3 iteration with improved MoE architecture

Context: 128K
Input: $0.27/1M
Output: $1.10/1M
textcodefunction-calling

DeepSeek V3.1

✅ Active

Enhanced V3 with better instruction following and reduced hallucinations

Context: 128K
Input: $0.27/1M
Output: $1.10/1M
textcodefunction-calling

DeepSeek V3

✅ Active

671B MoE model trained for $5.6M, competing with GPT-4 class models

Context: 128K
Input: $0.27/1M
Output: $1.10/1M
textcodefunction-calling

DeepSeek R1

✅ Active

Open-source reasoning model competitive with o1 at fraction of the cost

Context: 128K
Input: $0.55/1M
Output: $2.19/1M
textreasoningcode

DeepSeek R1-0528

✅ Active

Updated R1 with improved math and coding performance

Context: 128K
Input: $0.55/1M
Output: $2.19/1M
textreasoningcode

Change History

Feb 1, 2026New ReleaseDeepSeek
DeepSeek V3.2

DeepSeek V3.2 released with improved reasoning and efficiency

Sep 15, 2025New ReleaseDeepSeek
DeepSeek V3.1

DeepSeek V3.1 released with improved instruction following

May 28, 2025New ReleaseDeepSeek
DeepSeek R1-0528

R1-0528 released with significant improvements in math reasoning

Feb 15, 2025UpdateDeepSeek
DeepSeek R1

R1 API pricing reduced significantly

Jan 20, 2025UpdateDeepSeek
DeepSeek V3

V3 API generally available with 128K context

Jan 20, 2025New ReleaseDeepSeek
DeepSeek R1

DeepSeek R1 released as open-source reasoning model, rivaling o1

Dec 26, 2024New ReleaseDeepSeek
DeepSeek V3

DeepSeek V3 launched, trained for just $5.6M