TellWorldAI World Archive
Ask中文
News
Models
Organisations
People
Papers & concepts
Products & agents
Chips & robots
History
Topics
Ask中文

Home›Models

Model

DeepSeek-V3.1

DeepSeek-V3.1, released on 21 August 2025 under the MIT License, was DeepSeek's first hybrid model, supporting both a thinking mode and a non-thinking mode in one set of weights. It has 671 billion total parameters, with 37 billion activated per token.

Status DeprecatedModality TextOpen weights YesContext (K tokens) 128 Era Generative AI for everyone

Parameters
671B
License
MIT
Access
API, App, Download
Release date
Developer
DeepSeek

What it is

DeepSeek-V3.1 is a language model that DeepSeek released on 21 August 2025 under the MIT License. [1][2]

It was DeepSeek's first hybrid model. One model supports both thinking and non-thinking modes. In the API, the name deepseek-chat maps to non-thinking mode and deepseek-reasoner maps to thinking mode. [1]

The model has 671 billion total parameters, 37 billion of them activated, and a 128K context in both modes. [2]

A hybrid model can switch between two ways of answering. In thinking mode it works through intermediate steps before responding, which usually takes longer and tends to suit harder problems. In non-thinking mode it answers directly, which is faster. The same weights serve both modes, so developers do not need to run two separate models.

Origins

DeepSeek-V3.1-Base extends the V3 base model with two long-context training phases. The first used 630 billion tokens at 32K, ten times more than before, and the second used 209 billion tokens at 128K, 3.3 times more. [2]

The model was trained with the UE8M0 FP8 scale data format for both weights and activations, for compatibility with microscaling data formats. [2]

Main features

In non-thinking mode, the model scored 66.0 on SWE-bench Verified (agent), against 45.4 for V3-0324, and 31.3 on Terminal-bench, against 13.3. [2]

In thinking mode, it scored 88.4 on AIME 2025, 80.1 on GPQA-Diamond, 74.8 on LiveCodeBench and 30.0 on BrowseComp. For comparison, DeepSeek-R1-0528 scored 87.5, 81.0, 73.3 and 8.9 on the same tests. [2]

DeepSeek said the thinking mode reached answers faster than DeepSeek-R1-0528 and was stronger at multi-step search-agent tasks. [1]

New API pricing took effect on 5 September 2025 at 16:00 UTC, ending DeepSeek's off-peak discounts. [1]

Its place in development

Its update, V3.1-Terminus, followed on 22 September 2025. The weights were open-sourced on Hugging Face, and about a week later Terminus became the base for the experimental DeepSeek-V3.2-Exp. [3][4]

CapabilitiesLarge language modelsReasoningAgents & tool use

Ecosystem & regionOpen sourceChina

Version history

12
  1. 2023
    • Nov 29, 2023 DeepSeek LLM (7B and 67B)

      DeepSeek LLM is DeepSeek's first general-purpose language model series, released on 29 November 2023 in 7B and 67B sizes, each with a base version and a chat version.

  2. 2024
    • May 7, 2024 DeepSeek-V2

      DeepSeek-V2 is a mixture-of-experts language model released in May 2024 with 236B total parameters, of which 21B are activated per token. It introduced Multi-head Latent Attention and used DeepSeekMoE for economical sparse training.

    • Sep 5, 2024 DeepSeek-V2.5

      DeepSeek-V2.5 is a model released on 5 September 2024 that merged DeepSeek's chat and code models into one, keeping both conversational and coding ability while improving safety scores.

    • Dec 26, 2024 DeepSeek-V3

      DeepSeek-V3 is an open-weight mixture-of-experts language model that DeepSeek released in December 2024. Each token uses 37 billion of its 671 billion parameters, and DeepSeek describes its chat model as comparable to leading closed-source systems.

  3. 2025
    • Jan 20, 2025 DeepSeek-R1

      DeepSeek-R1 is an open-weight reasoning model that DeepSeek released on 20 January 2025. Built on DeepSeek-V3-Base and trained with reinforcement learning, it is released under the MIT License, which allows commercial use and distillation.

    • Mar 24, 2025 DeepSeek-V3-0324

      DeepSeek-V3-0324 is a March 2025 update to DeepSeek-V3 that keeps the same model structure while raising benchmark scores and moving to the MIT License.

    • May 28, 2025 DeepSeek-R1-0528

      An updated release of DeepSeek's open-weight reasoning model, published on 28 May 2025 under the MIT License, with markedly higher scores on competition math, science and coding benchmarks, plus system-prompt support and stronger function calling.

    • Aug 21, 2025 DeepSeek-V3.1 · This page

      DeepSeek-V3.1, released on 21 August 2025 under the MIT License, was DeepSeek's first hybrid model, supporting both a thinking mode and a non-thinking mode in one set of weights. It has 671 billion total parameters, with 37 billion activated per token.

    • Sep 29, 2025 DeepSeek-V3.2-Exp

      An experimental DeepSeek model released on 29 September 2025 that introduced DeepSeek Sparse Attention to make long-context training and inference cheaper, shipped with open weights and an immediate API price cut.

    • Dec 1, 2025 DeepSeek-V3.2 and V3.2-Speciale

      DeepSeek-V3.2, released on 1 December 2025, succeeded the experimental V3.2-Exp and integrated thinking into tool use. Its companion V3.2-Speciale, which maximizes reasoning, reached gold-level results at several 2025 competitions.

  4. 2026
    • Apr 24, 2026 DeepSeek-V4 (V4-Pro and V4-Flash)

      DeepSeek-V4, released in preview on 24 April 2026, comprises V4-Pro and V4-Flash, two open mixture-of-experts models with a 1M-token context. Its architecture targets long-context efficiency, and its later releases extended the series with agent upgrades and multimodal input.

    • Sep 10, 2026 DeepSeek-V4.1-Flash

      DeepSeek-V4.1-Flash, released on 10 September 2026 under the API name deepseek-flash, is a 552-billion-parameter multimodal mixture-of-experts model with a causal encoder-decoder design and a 1M-token context.

20253

  1. UpdateDeepSeek releases the DeepSeek-V3.1-Terminus update

    On 22 September 2025 DeepSeek released DeepSeek-V3.1-Terminus, an update to V3.1 that reduced Chinese-English mixing and stray characters and strengthened its Code Agent and Search Agent.

  2. ReleaseZhipu releases GLM-4.6 with a 200K context and domestic-chip support

    In late September 2025 Zhipu released GLM-4.6 under the MIT License, with a 200K-token context window, stronger coding and agent abilities, and support for Chinese domestic chips.

  3. ReleaseDeepSeek releases hybrid-thinking DeepSeek-V3.1

    On 21 August 2025 DeepSeek released DeepSeek-V3.1, its first hybrid model, which runs in either a thinking or a non-thinking mode under the MIT License.

Developed by1

DeepSeek

Successor of2

DeepSeek-R1-0528DeepSeek-V3-0324

Succeeded by1

DeepSeek-V3.2-Exp

Powers1

DeepSeek app

Uses2

Chain of thought and reasoning modelsMixture of experts (MoE)

Notes and sources

  1. DeepSeek V3.1 Releasehttps://api-docs.deepseek.com/news/news250821
  2. deepseek-ai/DeepSeek-V3.1https://huggingface.co/deepseek-ai/DeepSeek-V3.1
  3. DeepSeek V3.1 Updatehttps://api-docs.deepseek.com/news/news250922
  4. DeepSeekhttps://en.wikipedia.org/wiki/DeepSeek

Cite this page: DeepSeek-V3.1. TellWorld, https://www.tellworld.net/models/deepseek-v3-1. Updated 2026-11-08