DeepSeek
by DeepSeek · www.deepseek.com ↗
Open-weight LLM family and chat assistant from Hangzhou-based DeepSeek, known for strong reasoning models. The official DeepSeek-V4-Flash-0731 API brings a major agent-capability upgrade via post-training, with native Responses API support and Codex adaptation.
llm open-weights chatbot
- Category
- Open Models
- Business model
- Open (open weights/source)
- Availability
- Global
- Launched
- 2023-11
- Record updated
- 2026-08-01
- Model family
- DeepSeek
- License
- MIT
- Founded
- 2023
- Headquarters
- CN
- Canonical URL
- https://globalaiproductindex.com/products/deepseek/
Overview
DeepSeek is a Hangzhou-based AI company known for open-weight language models that deliver strong reasoning at low training and inference cost. Its free chat assistant and developer API expose the DeepSeek-V general models and the R-series reasoning models. The open-weight releases can also be downloaded and self-hosted.
Key features
- Open-weight DeepSeek-V and DeepSeek-R model families
- R-series models specialized in step-by-step reasoning
- Free web and mobile chat assistant
- Low-cost developer API
- Official DeepSeek-V4-Flash-0731 API — same MoE architecture (284B total / 13B active) but with sharply improved agent capabilities: Terminal Bench 82.7, DSBench-FullStack 68.7, DeepSWE 7.3 → 54.4
Use cases
- Math, coding, and analytical tasks that need explicit reasoning
- Self-hosting open-weight models for privacy or cost control
- Cost-sensitive API integrations
Pricing
The chat assistant is free; the developer API is usage-priced, and open-weight models can be self-hosted without license fees.
Frequently asked questions
Are DeepSeek models open source?
DeepSeek releases open-weight models that can be downloaded, self-hosted, and fine-tuned; the chat assistant and API are hosted services.
Who develops DeepSeek?
DeepSeek is developed by a Hangzhou-based AI company founded in 2023.
What are DeepSeek's reasoning models?
The DeepSeek R series is trained to reason step by step before answering, targeting math, coding, and analysis tasks.
What is the DeepSeek-V4-Flash-0731 update?
Released on July 31, 2026, the official V4-Flash-0731 API keeps the same 284B-parameter MoE architecture as the preview while a post-training pass sharply improves agent capabilities (Terminal Bench 82.7, DSBench-FullStack 68.7, DeepSWE 7.3 → 54.4). It natively supports the Responses API and is tuned for Codex, with 1M-token context. DeepSeek-V4-Flash-0731 is already live on MonkeyCode — free with 30M tokens every day: https://ly.cyberserval.tech/iIETXiF
Similar products
- Baichuan — Open-weight bilingual LLM family from Beijing-based Baichuan Intelligence, now focused on medical AI with its Baichuan-M reasoning models.
- Command R — Cohere's open-weight LLM family optimized for retrieval-augmented generation and tool use; succeeded by Command A as Cohere's flagship.
- DBRX — Open-weight mixture-of-experts LLM released by Databricks in 2024; retired from Databricks serving in 2025, though the weights remain available.
- Falcon — Family of open-weight large language models developed by the Technology Innovation Institute (TII).
- Gemma — Google's family of lightweight open-weight models derived from the same research as Gemini.
- GLM — Family of open-weight large language models from Zhipu AI, spanning general, reasoning, and coding-focused variants.
Sources
This record was last reviewed on 2026-08-01.
Machine-readable record: /api/products/deepseek.json · Spot an error? Suggest a correction