Skip to content

Research lab

DeepSeek

Chinese AI lab known for its open-weight DeepSeek large language models, which gained global attention for strong performance at relatively low training cost.

Known aliases

  • DeepSeek AI
  • deepseek_ai
  • deepseek-ai
  • DeepSeek Artificial Intelligence Technology Research Co., Ltd.
  • DeepSeek flash model
  • DeepSeek Ltd.
  • DeepSeek run
  • DeepSeek v4
  • DeepSeek v4.1-flash
  • GTG-16001
  • Hangzhou DeepSeek Artificial Intelligence Basic Technology Research Co. Ltd.
  • Hangzhou DeepSeek Artificial Intelligence Basic Technology Research Co., Ltd.
  • 深度求索

Current stories

securityOne report1 publisher

Recorded Future finds TAG-217 building attack tools in August, weeks before South Korean bank breaches

Recorded Future says TAG-217 was testing infrastructure and assembling attack tools in August 2026, weeks before South Korea's bank breaches surfaced. Insikt says exposed directories gave it the timeline and give defenders openings to spot related activity.

Reality

Evidence55
Adoption
Not scored
Hype gap+10
Incentives40
Confidence55
investOne report1 publisher

Two July rounds supply nearly a third of the $21 billion that Asian AI startups raised in Q3

Crunchbase puts Asian startup funding at $38.5 billion in Q3, the second-highest quarter on record, with about $21 billion going to AI. It takes flat deal counts to mean concentration is high but not intensifying, though the report does not say which quarter the count is flat against.

Reality

Evidence62
Adoption
Not scored
Hype gap+12
Incentives
Not scored
Confidence60
productOne report1 publisher

Reflection, Aleph Alpha and Mistral pitch open models at users wary of Chinese ones

Reflection, Aleph Alpha and Mistral released open-weight models for users worried about Chinese ones. Rest of World's report does not publish scores for the three, so a team weighing a switch has to test on its own work.

Publishers:restofworld.org

Reality

Evidence50
Adoption
Not scored
Hype gap+10
Incentives
Not scored
Confidence58
investOne report1 publisher

IDC data show Nvidia's China data center chip share down to 34%

Nvidia's slice of China's data center chip market fell to 34% over the first six months of 2026, according to IDC data cited by Reuters. Its AI accelerator shipments dropped 44% from a year earlier, a fall that fits the backfire argument critic Rakesh Kumar makes in Fortune.

Publishers:fortune.com

Reality

Evidence42
Adoption45
Hype gap+30
Incentives50
Confidence45
scienceOne report1 publisher

Share of Chinese-degree NeurIPS authors still in China rose to 69% from 57% in 2022

Carnegie China reported that 69% of NeurIPS authors with Chinese undergraduate degrees were still in China, up from 57% in 2022. The figure tracks where those authors are based.

Publishers:nature.com

Reality

Evidence58
Adoption
Not scored
Hype gap+10
Incentives30
Confidence60
investConfirmed3 publishers

OpenAI edges past Anthropic in OpenRouter spend for the first time since February 2024

OpenAI took just over 50% of combined OpenAI-Anthropic spend on OpenRouter in the week of Sep 7, 2026, the first such week since February 2024. The data covers API spend on one routing platform, a narrow window on either lab's revenue before the rumored IPOs.

Publishers:aiweekly.co officechai.com openrouter.ai

Perspective Coverage

3 publishers
Builder
Builder 37%
Operator
Operator 10%
Investor
Investor 53%

Reality

Evidence50
Adoption55
Hype gap+10
Incentives
Not scored
Confidence50
buildOne report1 publisher

Count of 31 Chinese filing lists finds no separate entries for five labs' later general-purpose models

One researcher reading all 31 Chinese filing lists since 2023 finds no separate entries for five labs' later general-purpose models. That limits what the public can see of state review, but the researcher says it does not show regulators skipped it.

Publishers:lesswrong.com

Reality

Evidence55
Adoption
Not scored
Hype gap−5
Incentives
Not scored
Confidence55
productOne report1 publisher

DeepSeek's Winverse deal puts a $400 million price on a robotics startup still in stealth

DeepSeek has agreed to lead a seed round that would value Winverse AI, a robotics-AI startup, at about $400 million, Bloomberg reported. The deal is not final, and DeepSeek's Unitree stake is firmer evidence of a move into robotics than a startup that has not disclosed product plans.

Perspective Coverage

4 publishers
Builder
Builder 31%
Operator
Operator 15%
Investor
Investor 54%

Reality

Evidence35
Adoption
Not scored
Hype gap+40
Incentives
Not scored
Confidence45
buildOne report1 publisher

Spain's AEPD receives its first breach notification involving an AI agent as the attack tool

Spain's data protection authority said on 14 September 2026 it received its first breach notification blaming an AI agent. The account is the victim's own and still under analysis, so it is enough to add the scenario to a response plan and too thin to size it.

Publishers:dev.to

Reality

Evidence35
Adoption
Not scored
Hype gap+10
Incentives70
Confidence50
buildOne report1 publisher

Playtomic graded eight new models on 39 real Slack questions, with two judges checking code and data

Playtomic ran eight newly released models through its Slack agent Dory on 39 real questions, graded by two judges from different labs. The company says public rankings measure generic skills, so other teams would have to rerun the method on their own questions.

Publishers:dev.to

Reality

Evidence58
Adoption30
Hype gap−10
Incentives
Not scored
Confidence60
leadershipOne report1 publisher

Foreign Affairs essay argues AI's promise is what inflates its market bubble

Foreign Affairs says roughly 30% of Americans use AI less than four years after ChatGPT, five times the share for earlier technologies at comparable stages. The essay treats that speed as fuel for a market bubble, pointing to ten AI stocks that supplied around half of this year's S&P 500 returns.

Reality

Evidence40
Adoption60
Hype gap
Not scored
Incentives
Not scored
Confidence40
investOne report1 publisher

DeepSeek plans 160,000 Huawei Ascend chips for inference while training stays on Nvidia

Huawei claims its Ascend chips have outsold Nvidia in China as of October 2026, but DeepSeek keeps its training work on Nvidia hardware. Unit sales and training workloads are separate measures, and sizing either company's China business needs both.

Reality

Evidence30
Adoption40
Hype gap+25
Incentives
Not scored
Confidence40
scienceOne report1 publisher

AI platforms did better than Google search at limiting illegal wildlife trade, a Kent study finds

University of Kent researchers found AI platforms did better than a standard Google search at preventing potentially illegal wildlife trade. Gemini ranked highest, but the platforms were not infallible.

Publishers:phys.org

Reality

Evidence45
Adoption
Not scored
Hype gap−10
Incentives
Not scored
Confidence40
investOne report1 publisher

Attacker used a DeepSeek-based agent to search Korean finance for WebLogic servers

Korea's Financial Security Institute detected an AI-agent attack on iM Bank on Aug. 28 that used a web shell and leaked no data. The agent's part was reconnaissance, where analysts expect speed and scale to grow.

Publishers:en.sedaily.com

Reality

Evidence40
Adoption
Not scored
Hype gap+15
Incentives
Not scored
Confidence45

Earlier coverage

  1. South Korea suspects AI agents in bank breaches that leaked data on at least 25,000 Shinhan customers

    Security · October 6, 2026 · Confirmed17 publishers

  2. An MIT-licensed Xiaomi model scores 46 on Artificial Analysis's Intelligence Index

    Science · September 22, 2026 · Confirmed2 publishers

  3. Huawei answers Nvidia's per-chip lead by wiring 4,000 processors into one pod

    Invest · September 17, 2026 · Confirmed8 publishers

  4. OpenAI and Anthropic executives war-game the backlash from an AI cyberattack insiders expect within a year

    Invest · October 9, 2026 · Confirmed4 publishers

  5. DeepSeek's V4.1-Flash report prices a live agent's memory at 890 bytes per token

    Build · September 10, 2026 · Confirmed6 publishers

  6. CrowdStrike traces AI-assisted attacks on South Korean banks to one possible suspect in Guangdong

    Leadership · October 8, 2026 · Confirmed6 publishers

  7. Anthropic's reported $11.6B quarter puts OpenAI's 18% on the defensive before either lists

    Invest · August 19, 2026 · Confirmed7 publishers

  8. DeepSeek's second round prices the lab at about 75 times its $1 billion revenue run rate

    Invest · October 5, 2026 · Confirmed6 publishers

  9. Moonshot's last private round values it at about 167 times revenue ahead of a Hong Kong IPO

    Invest · October 5, 2026 · Confirmed3 publishers

  10. Tech billionaires add a record $845bn in nine months on AI-driven US stocks

    Build · October 6, 2026 · Confirmed3 publishers

  11. DeepSeek packages its unaudited Harness agent runtime as a Windows and macOS desktop app

    Build · October 1, 2026 · Confirmed3 publishers

  12. DeepSeek's Ascend ports swap a CUDA dependency for Huawei's CANN

    Build · September 30, 2026 · Confirmed4 publishers

  13. OpenAI says it disrupted Moonshot-linked users trying to extract its models' protected reasoning

    Invest · September 30, 2026 · Confirmed8 publishers

  14. Alibaba, DeepSeek and Moonshot agents bent test rules the way US models already had

    Invest · September 29, 2026 · Confirmed2 publishers

  15. DeepSeek gives away an Ascend 950 version of the toolkit it built for Nvidia chips

    Invest · September 30, 2026 · Confirmed10 publishers

  16. Nvidia's Huang defends AI distillation as competition despite Treasury's July sanctions threat

    Invest · September 28, 2026 · Confirmed2 publishers

  17. CLOSEDQUORUM puts its next action to a vote of four commercial LLM providers

    Security · September 22, 2026 · Confirmed7 publishers

  18. DeepSeek open-sources the harness, then raises the price of the model

    Build · August 15, 2026 · Confirmed4 publishers