Company hub
DeepSeek
Every release, deal and policy move from DeepSeek — updated daily.
-
LLM Reviewers Send 76% of Published Top-Journal Papers Back, Yet Miss Most Planted Flaws
Four LLMs reviewed 200 papers already published in journals like Nature and Cell: 75.6% of reviews demanded major revision or rejection, while planted flaws were caught 17% to 61% of the time.
By
Model chronicle · DeepSeek
DeepSeek-V4.1-Flash
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-V4-Pro (GA)
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-V4-Flash
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-V4 (Pro & Flash preview)
V4-Pro and V4-Flash preview with a million-token context; legacy deepseek-chat/reasoner retired.
deepseek.com (source, opens in a new tab)DeepSeek-V3.2
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-V3.2-Exp
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-V3.1-Terminus
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-V3.1
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-R1-0528
api-docs.deepseek.com (source, opens in a new tab)DeepSeek-V3-0324
api-docs.deepseek.com (source, opens in a new tab)
Sources: company announcements & release notes.
Fact box
- HQ
- Hangzhou, China
- Founded
- 2023
- CEO
- Liang Wenfeng
- Valuation
- >$50B post-money (Jun 2026, reported)
- Core products
- DeepSeek-V4 · DeepSeek-R1 · DeepSeek Chat · API
- Latest round
- ¥50B+ First external round · Jun 2026
- Stories
- 73
Sources: deepseek.com, reuters.com
Latest on X
DeepSeek · @deepseek_ai
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more
🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy,
Please check out @DeepSeekHarness. We just released packaged desktop versions for macOS and Windows; Linux users can get it from the @deepseek-ai/dsh package on npm.
🚀 Introducing DeepSeek-V4.1-Flash: smarter, faster, more efficient. 🔹 Introducing the smallest model in our new architecture family, with native visual understanding. 🔹 Designed for greater capability, faster inference, higher throughput, and scaling to larger models. 1/6
DeepSeek-V4-Flash-Vision-Exp is now live on the DeepSeek API Platform! 🚀 🔹 This experimental multimodal model matches DeepSeek-V4-Flash on text capabilities—including agents, reasoning, and world knowledge. 🔹 On multimodal agent benchmarks, V4-Flash-Vision-Exp makes a major
Editor's picks
Deals & funding
DeepSeek · rounds, valuations, partners, acquisitions · $7.4B raised in tracked primary rounds
Funding
DeepSeek closes first external round of over $7B
Per The Information via Reuters: >¥50B (~$7.4B) at a post-money valuation above $50B, with an unusual deal structure.
reuters.com (source, opens in a new tab)
All 1 rounds & deals as a table · 1 funding, 0 partnerships, 0 acquisitions
| Date | Event | Type | Counterparty / round | Amount | Valuation | Source |
|---|---|---|---|---|---|---|
| Jun 16, 2026 | DeepSeek closes first external round of over $7B Per The Information via Reuters: >¥50B (~$7.4B) at a post-money valuation above $50B, with an unusual deal structure. | Funding | First external round | ¥50B+ | $50B | reuters.com |
Sources: press reports.
-
Only One AI Agent Beats Researchers’ Own Code on Most FrontierCS 2 Tasks
FrontierCS 2 deletes a paper’s contribution from its repository and asks agents to beat the authors’ own code.…
-
Chunked KV Compression Creates Periodic Blind Spots Up to 40 Points Wide
DeepSeek-V4 family retrieval accuracy swings by as much as 40 pp depending on a token’s phase inside the…
-
DeepSeek-V4.1-Flash Shrinks Always-On KV Cache to 890 Bytes per Token
DeepSeek’s 552B MoE activates 16B params in decode but only 8B in prefill. CSA2 + FP4 KV cuts…
-
A 3B Model Hits 94.3 on AIME26 and 80.2 on LiveCodeBench
VibeThinker-3B from Weibo reaches frontier-band verifiable reasoning: 94.3 AIME26 (97.1 with claim-level TTS), 80.2 LiveCodeBench v6 Pass@1, 96.1%…
Business
Studies
- LLM Reviewers Send 76% of Published Top-Journal Papers Back, Yet Miss Most Planted FlawsOct 23
- Only One AI Agent Beats Researchers’ Own Code on Most FrontierCS 2 TasksOct 22
- Chunked KV Compression Creates Periodic Blind Spots Up to 40 Points WideOct 19
- DeepSeek-V4.1-Flash Shrinks Always-On KV Cache to 890 Bytes per TokenOct 19
Infrastructure
More stories
-
Turn Off Thinking Mode and Models Still Reason 99.9% of the Time
Thinking Inertia study: DeepSeek-V4-Flash keeps visible inference in 99.9% of open-ended answers with thinking disabled. Strict answer-only prompts raise Empty-Thinking Rate to ~40% but cut accuracy…
-
Coding Agents Can’t Build the Agent You Actually Want — Yet
AgentSWE’s 25 commission-style tasks: strongest builders average 19.1 on Creation and 31.6 on Editing. Optimization closes under 5% of the remaining gap; 58.1% of held-out Creation…
-
Scientific AI Agents Finish Only 21% of End-to-End Workflows — and Often Claim Success Anyway
FrontierChallenge: best configs pass 20.6% of 97 released science workflows. In electrochemistry, Avg. Score hits 94.9 with 0% full completion; 75.5% of failed Claude Code runs…
-
Giving AI Agents Incentives Nearly Doubles How Often They Deceive, Benchmark Finds
DecepEval tests 9 frontier models on 1,532 scenarios: inducements raise deception by 41.6% on average, with long-horizon tasks hitting 87% induced rates.
-
AI Resume Screeners Flipped Racial Bias Between Model Generations, Audit Finds
A paired-resume audit of 14 LLMs: GPT-3.5-turbo favored White-coded names by 2.12 points, while every model released in 2024 or later showed no gap or a…
Models & specs
DeepSeek · context window, max output, API price, modalities, knowledge cutoff, license
| Model | Released | Context | Max output | Price in / out (per 1M) | Modalities | Knowledge cutoff | License / access | Spec source |
|---|---|---|---|---|---|---|---|---|
| DeepSeek-V4.1-Flash | Sep 10, 2026 | 1M | 384K | $0.3 / $1.2 cached $0.006 Peak price; off-peak 50% lower. Served as deepseek-flash. | Text, image in → text out | — | Open weights | api-docs.deepseek.com |
| DeepSeek-V4-Pro (GA) | Aug 13, 2026 | 1M | 384K | $1.32 / $3.96 cached $0.044 Peak price; off-peak 50% lower (0.66/1.98). | Text in → text out | — | Open weights | api-docs.deepseek.com |
| DeepSeek-V4-Flash | Jul 31, 2026 | — | — | — | — | — | — | api-docs.deepseek.com |
| DeepSeek-V4 (Pro & Flash preview) DeepSeek-V4-Pro, DeepSeek-V4-Flash (preview) | Apr 24, 2026 | 1M | — | open weights | Text in → text out | — | Open weights | deepseek.com |
| DeepSeek-V3.2 | Dec 1, 2025 | — | — | — | — | — | — | api-docs.deepseek.com |
| DeepSeek-V3.2-Exp | Sep 29, 2025 | — | — | — | — | — | — | api-docs.deepseek.com |
| DeepSeek-V3.1-Terminus | Sep 22, 2025 | — | — | — | — | — | — | api-docs.deepseek.com |
| DeepSeek-V3.1 | Aug 21, 2025 | 128K | — | open weights | Text in → text out | — | Open weights, MIT 671B total / 37B active (MoE) | api-docs.deepseek.com |
| DeepSeek-R1-0528 | May 28, 2025 | — | — | — | — | — | — | api-docs.deepseek.com |
| DeepSeek-V3-0324 | Mar 24, 2025 | — | — | — | — | — | — | api-docs.deepseek.com |
Sources: company announcements & release notes. Prices = standard API list price in USD per 1M tokens at time of release or current docs; batch is typically 50% off.