DeepSeek Timeline: Model Release Dates and Key Milestones

DeepSeek is a Chinese AI company founded in Hangzhou in 2023. Its first public model family, DeepSeek Coder, arrived on November 2, 2023. The company then expanded into general language models, mathematical reasoning, formal proof, vision, document recognition, and dedicated reasoning models.

DeepSeek-V4-Flash-Vision-Exp is the company’s newest model as of August 21, 2026. It is an experimental multimodal API model called through deepseek-v4-flash-vision-exp. It matches V4-Flash on text capabilities and accepts image input alongside text through the Chat Completions, Messages, and Responses APIs.

Between DeepSeek Coder and the V4-Flash-Vision-Exp update, the company repeatedly folded ideas from its specialist research into its main model line. It also expanded its developer platform with DeepSeek Harness, an open-source system for building and running AI agents. The dates below place these model, product, and API milestones in order.

Last Updated: August 21, 2026

Key DeepSeek Release Dates

DeepSeek Release History

Current DeepSeek Model Lineup

DeepSeek-V4-Pro-0813 is the current Pro model and is called through deepseek-v4-pro. It supports thinking and non-thinking modes within a one-million-token context window and can return up to 384K output tokens. The API supports JSON output, tool calls, the Responses API, the Anthropic API, chat prefix completion, and non-thinking FIM completion.

DeepSeek-V4-Flash-0731 is the current Flash version under the deepseek-v4-flash name. It has 284 billion total parameters and activates 13 billion for each token. DeepSeek re-post-trained the preview model for agent tasks without changing its architecture or size. Like Pro, it supports a 1M context window, dual reasoning modes, the Responses API, and the Anthropic API.

DeepSeek-V4-Flash-Vision-Exp is an experimental multimodal API model under the deepseek-v4-flash-vision-exp name. It matches V4-Flash on text capabilities while accepting mixed text and image input through Chat Completions, the Anthropic-compatible Messages API, and the Responses API. Images can be supplied as base64 data, external URLs, or Files API file IDs, and each image uses up to 384 billed input tokens at V4-Flash rates.

DeepSeek-V3.2 was the previous general flagship. Its open weights remain available, but V4 has replaced it on DeepSeek’s main API routes. V3.2 added reasoning during tool use and completed the architecture first tested in V3.2-Exp.

DeepSeek Harness and API Pricing

DeepSeek Harness v0.1, also called dsh, is an open-source agent harness built on the Cordis meta-framework. Models, tools, skills, sessions, sandboxes, filesystems, agent loops, orchestration, and the user interface are implemented as replaceable plugins. The August 13 release is a developer preview under the MIT License, and DeepSeek warns that compatibility-breaking changes will occur during development. Harness 0.1.1 added out-of-the-box support for V4-Flash-Vision-Exp on August 21, 2026.

DeepSeek’s API now uses peak and off-peak rates. Time-based billing began at 16:00 UTC on August 16, 2026, with off-peak prices set at half the peak rates. Peak hours are 01:00–04:00 and 06:00–10:00 UTC. For every one million tokens, V4-Flash will cost $0.014 for a cache-hit input, $0.44 for a cache-miss input, and $1.32 for output during peak hours. V4-Pro peak rates will be $0.044, $1.32, and $3.96 respectively. Images sent to V4-Flash-Vision-Exp use the V4-Flash rates after conversion to input tokens.

DeepSeek Timeline

2026: Vision, OCR-2, DeepSeek-V4, Harness, and API Changes

2025: R1, Hybrid Reasoning, Agents, and Specialist Models

2024: MoE Research, V2, V3, and the First Reasoning Preview

2023: DeepSeek Begins with Code and General Language Models

DeepSeek Model Families Explained

V Series

The V series is DeepSeek’s main general-purpose model line. V2 introduced the MoE and attention architecture that shaped later releases. V2.5 combined general and coding abilities, V3 increased the scale, V3.1 merged thinking with regular response modes, and V3.2 brought reasoning into tool use. V4 divided the main line into Pro and Flash models; Flash became the first to receive an official API release, and V4-Flash-Vision-Exp later added experimental image input to that API family.

R Series

The R series focuses on reasoning. R1-Lite-Preview was the first hosted preview. R1-Zero tested reinforcement learning without supervised fine-tuning as the first post-training stage, while R1 added cold-start data to improve readability and consistency. The six distilled R1 models transferred parts of that reasoning behavior into smaller Qwen- and Llama-based models.

Coder, Math, and Prover

DeepSeek Coder and Coder-V2 were built for programming. DeepSeekMath concentrated on competition-style mathematical problems expressed in natural language. DeepSeek-Prover targeted formal proofs that a Lean proof assistant could check. These branches influenced the general models even when their names disappeared from the main API.

VL, Janus, and OCR

DeepSeek-VL and VL2 handle vision-language understanding. Janus combines image understanding with image generation, but uses different visual encoding paths for the two jobs. DeepSeek-OCR treats document images as compressed carriers of text and layout. The newer DeepSeek-V4-Flash-Vision-Exp adds an experimental multimodal API branch that accepts images with text through Chat Completions, Messages, and Responses. It remains a separate vision variant alongside the general V4-Pro and V4-Flash endpoints.

What the Model Suffixes Mean

Five Turning Points in DeepSeek’s History

1. The First Releases Were Specialist Models

DeepSeek did not begin with a consumer chatbot. It released a code model first, then a general language model, followed by separate research families for MoE routing, mathematics, and vision. Each branch tested an idea that could later move into a larger general model.

The specialist work also fed later releases. DeepSeekMath explored reinforcement-learning methods used in R1. DeepSeekMoE supplied the sparse expert design behind V2 and V3. Coder and Math data helped train Coder-V2 and the general models that followed.

2. V2 Established the Efficiency Architecture

V2 was the first general DeepSeek model built around the architecture now associated with the company. Its mixture-of-experts system activated only part of the full parameter set for each token. Multi-head Latent Attention reduced the memory required for the key-value cache during inference.

Those choices carried into later releases. V3 used a much larger MoE design, R1 inherited the V3 base architecture, and V4 continued the effort to lower the cost of long-context inference. V2 is the architectural starting point for the later flagships.

3. V3 Became the General Foundation

V3 expanded the MoE design to 671 billion total parameters while activating 37 billion for each token. It handled general writing, coding, mathematics, and other text tasks through one model. The DeepSeek App used V3 at launch, and the official deepseek-chat API identifier pointed to V3 at the time.

V3 also became the base for R1. R1 was not a separate architecture built from the ground up. It applied a reasoning-focused post-training process to the capabilities and efficiency of the V3 foundation.

4. R1 Took DeepSeek Beyond Its Developer Audience

R1 made DeepSeek widely known in January 2025. The release combined a large reasoning model, an experimental R1-Zero checkpoint, and six smaller distilled models. Developers could inspect and run the weights, while ordinary users could access reasoning through the DeepSeek service.

The App had launched five days earlier with V3. R1 expanded what users could do inside the existing product, but it was not the App’s launch model. The later R1-0528 update added JSON output and function calling for software integrations.

5. V3.1 and V4 Merged Reasoning with Agent Work

V3.1 reduced the separation between a normal chat model and a reasoning model. One model could answer in thinking or non-thinking mode, and the API identifiers selected the mode. V3.2 extended that design by allowing the model to reason while using tools.

V4 kept both modes and gave Pro and Flash a 1M context window. The July 2026 Flash update strengthened multi-step agent work and added native Responses API support, which allowed Codex clients to call the model through DeepSeek’s API. V4-Flash-Vision-Exp extended that API path to mixed text-and-image requests on August 21, 2026.

Read More: How to use DeepSeek v4 Pro & Flash in Codex and Cut API costs

DeepSeek Timeline FAQs

When was DeepSeek founded?

DeepSeek was founded in Hangzhou, China, in 2023 by Liang Wenfeng. High-Flyer, the quantitative investment firm he co-founded, provided financial backing. No single founding day is consistently documented because the research lab’s creation and its later corporate separation occurred at different stages.

What was the first DeepSeek model?

DeepSeek Coder was the first public DeepSeek model family. It was released on November 2, 2023, with models for code completion and instruction-based programming tasks. DeepSeek LLM followed on November 29 as the company’s first general-purpose language model family.

When was DeepSeek-R1 released?

DeepSeek-R1 was released on January 20, 2025. The release included DeepSeek-R1-Zero, the main R1 model, and six distilled models based on Qwen and Llama architectures. DeepSeek updated the reasoning family with R1-0528 on May 28, 2025.

Did the DeepSeek App launch with R1?

No. The official DeepSeek App launched on January 15, 2025, and was initially powered by DeepSeek-V3. R1 arrived on January 20 and was then made available through DeepSeek’s web product, App, and API. The two events happened in the same week but were separate releases.

What is the latest DeepSeek model?

DeepSeek-V4-Flash-Vision-Exp is the newest model announced as of August 21, 2026. It is an experimental multimodal API model served through deepseek-v4-flash-vision-exp, matches V4-Flash on text capabilities, and accepts image input through Chat Completions, Messages, and Responses. DeepSeek-V4-Pro-0813 remains the current Pro model, while DeepSeek-V4-Flash-0731 remains the general Flash option.

Does DeepSeek-V4-Flash-Vision-Exp support images?

Yes. The experimental model accepts JPEG, PNG, GIF, and WebP images alongside text. You can provide an image as base64 data, an external URL, or a Files API file_id through the Chat Completions, Messages, or Responses API. Images are converted to input tokens for billing, with a maximum of 384 tokens per image at V4-Flash rates.

What is the difference between DeepSeek V models and R models?

The V series is DeepSeek’s main general-purpose line. It covers ordinary chat, writing, coding, tool use, and other text tasks. The R series was created for deliberate reasoning, especially mathematics, coding, and multi-step problems. Starting with V3.1, the V series also gained thinking modes, so the boundary is less rigid than it was when R1 launched.

Is DeepSeek open source?

Most DeepSeek models have downloadable weights, technical reports, and permissive licenses. R1 and several recent V-series releases use the MIT License.

What do deepseek-chat and deepseek-reasoner mean?

deepseek-chat and deepseek-reasoner are legacy API compatibility identifiers whose underlying models changed over time. They most recently selected V4-Flash’s non-thinking and thinking modes.

Related Resources

• OpenAI & ChatGPT Timeline
• Anthropic Claude Timeline
• Google AI & Gemini Timeline
• Meta AI Timeline
• Reasonix: Free DeepSeek Coding Agent with Cache-First Cost Control
• CodeWhale (DeepSeek TUI): Free, Open-source Claude Code Alternative
• Open-Source Browser Automation & AI Assistant – DeepSeek Cowork
• DeepSeek Harness: Open-Source Plugin-Based AI Agent Harness
• Best DeepSeek Harness Plugins for DSH