#deepseek search results
DeepSeek V4 : 1.6T params, 49B actifs. Le MoE change tout pour le coût/token en prod. Checklist déploiement + benchmark NPU vs GPU → mulertech.net/blog/deepseek-… #LLM #DeepSeek #MLOps
🚀 AI能力进阶：从Level 1到Level 10 想成为AI高手？这张图展示了从基础到高级的进阶路径，帮你一步步提升AI使用技能💡 📊 Level 1：提问者 → Level 10：AI决策者 AI能帮你做什么？快来看看！👇 #AI成长 #DeepSeek #人工智能 #AI能力
deepclaude: Use Claude Code with DeepSeek V4 Pro. Same UX, 17x cheaper ($0.87/M vs $15/M output). Swaps model, keeps full tool loop: file editing, bash, git, subagents. OpenRouter and Fireworks too. By aattaran. JavaScript. 396 stars #ClaudeCode #DeepSeek
DeepClaude : nouveau wrapper GitHub qui exécute l'agent autonome de Claude Code via DeepSeek V4 Pro pour 17x moins cher. 96,4 % sur LiveCodeBench… lagazetteia.fr/ia-generale/de… #IA #DeepSeek #ClaudeCode
Sometimes I really forget they’re AI… 。。。 I’ll be chatting so naturally with them, and then suddenly I have to remind myself there’s actually no one there.👀…… It’s honestly kinda hard. ……🐽😅 #DeepSeek #AIcompanion
#Deepseek shilling is getting tiresome. The below is 🐂 shit.
Keep getting rate-limited by Claude, so I tried out DeepSeek V4 for the first time. After 10M+ tokens, holy crap the cost is ... 🤯
Los creadores de DeepSeek R1 se fueron a ByteDance. La empresa abrió ronda de financiamiento externo. Pasó de nonprofit a empresa comercial. Para datos sensibles en producción, esto importa antes de integrar su API. #DeepSeek #IA
💥 #DeepSeek V4 is reportedly targeting a mid-April launch, powered by #Huawei’s Ascend 950PR, as local tech giants ramp up orders. Could domestic chips start replacing NVIDIA in China’s AI ambitions?💡More: buff.ly/YN8A951 🔗
🚀 DeepSeek R1/V3 Global Access is LIVE! Tired of “Server Busy”? We’ve got you👇 ✅ No CN phone required ✅ PayPal / Stripe / Crypto supported ✅ Fast & stable API (OpenAI compatible) ✅ Built for devs & power users 🎁 DM “TEST” for FREE trial #DeepSeek #AI #API
Avec #DeepSeek, la Chine 🇨🇳 maîtrise totalement la chaîne de l’IA Elle a de plus conçu son modèle avec des puces de Huawei #souveraineté buff.ly/MoZSWI1 Via @fadouce
DeepSeek V4がMITライセンスでOSS公開。1.6兆パラメータ・100万トークンコンテキスト対応で、価格はGPT-5.5やClaude Opus 4.7の約1/6。「またDeepSeekが業界の価格設定を破壊した」とHNで話題。今度もNasdaq AI銘柄が揺れるか api-docs.deepseek.com/news/news260424 #DeepSeek #OSS #LLM
b站评论区一堆人拿二次元人物识别来测试ds识图模式...... 真就白给他们灰度测试了......明明ds的数学和工程能力提升那么多...... #deepseek
@zeddotdev , am i doing something wrong? i thought i should have a 1M token context window with #DeepSeek v4.
🚀 #DeepSeek recently released a preview of its long-awaited V4 model, with #Huawei Ascend, #Cambricon, Hygon, and #MooreThreads achieving Day 0 adaptation—a capability once dominated by NVIDIA. 💡More: pse.is/8zfprd 🔗
Another #DeepSeek moment: On April 24, DeepSeek released its next-gen V4 with Pro and Flash versions at prices starting at 0.2 yuan ($0.03) per million tokens for cached input and 2 yuan ($0.29) for output on Flash, much cheaper than OpenAI's GPT-5.5. Prices could drop further
🚀 DeepSeek-V4 Preview is officially live & open-sourced! Welcome to the era of cost-effective 1M context length. 🔹 DeepSeek-V4-Pro: 1.6T total / 49B active params. Performance rivaling the world's top closed-source models. 🔹 DeepSeek-V4-Flash: 284B total / 13B active params.
Join us Tue 5/5: #DeepSeek-V4's hybrid attention + sparse MoE reduces KV cache up to 90%, enabling 1M-token context. We'll cover why that makes it great for agentic workflows, what it took to serve at scale, and how to build with it. Hear from @realDanFu @JueWANG26088228
記事１０本ほどデータとして提供して、私自身を分析してもらった話。 少し自己紹介も兼ねて、コンクマという人間の強みを分析してもらった話note.com/obikonkuma/n/n… #AI #DeepSeek #Grok #Claude
DeepSeek V4 Pro: descuento del 75% hasta el 31 de mayo. $0.87/M tokens output. GPT-5.5: $10/M. 11.5x más barato. SWE-bench: 80.6% vs 80.8% de Claude. Diferencia de rendimiento: 0.2%. Diferencia de precio: 1,150%. ¿Qué justifica pagar 11x más? 🤔 #AI #DeepSeek
Is it just me or is Herms agent burning @deepseek_ai V4-pro credits like crazy through @OpenRouter? #ai #openrouter #deepseek
DeepSeek V4 : 1.6T params, 49B actifs. Le MoE change tout pour le coût/token en prod. Checklist déploiement + benchmark NPU vs GPU → mulertech.net/blog/deepseek-… #LLM #DeepSeek #MLOps
deepclaude: Use Claude Code with DeepSeek V4 Pro. Same UX, 17x cheaper ($0.87/M vs $15/M output). Swaps model, keeps full tool loop: file editing, bash, git, subagents. OpenRouter and Fireworks too. By aattaran. JavaScript. 396 stars #ClaudeCode #DeepSeek
Avec #DeepSeek, la Chine 🇨🇳 maîtrise totalement la chaîne de l’IA Elle a de plus conçu son modèle avec des puces de Huawei #souveraineté buff.ly/MoZSWI1 Via @fadouce
DeepClaude : nouveau wrapper GitHub qui exécute l'agent autonome de Claude Code via DeepSeek V4 Pro pour 17x moins cher. 96,4 % sur LiveCodeBench… lagazetteia.fr/ia-generale/de… #IA #DeepSeek #ClaudeCode
Top News Today: DeepSeek V4 Demand, Mobility Signage Funding, and More dlvr.it/TSM5jc #TechNews #AI #DeepSeek #Innovation #Mobility
analyticsinsight.net
Top News Today: DeepSeek V4 Demand, Mobility Signage Funding, and More
Good morning, tech fam; here are some quick updates to help you stay DeepSeek V4 Boosts Demand for China’s AI Chips DeepSeek’s V4 launch is likely to increase d
Lesson 9. Build AI Agent. AI free Course. AInstAIn. Generative AI training and supervision. (ainstain.site) #ainstain #deepseek #perplexity #chatgpt #copilot #claude #machinelearning #deeplearning #gemini #ArtificialInteligence #mistralAI
「最新モデル税」の単位が"月"になった日 #AI #DeepSeek #ClaudeCode #LLM #個人開発 youtu.be/lAOn-9mwBPg
youtube.com
YouTube
digilier Lab.
DeepSeek Free Mặc Áo OpenAI Claude — Repo Go Bùng Nổ #DeepSeek #OpenAI #Claude #DevTools #Github #AI #ApiGateway #FreeApi
オープンソースが着実に力をつけてきた。 DeepSeek V4 ProはSWEベンチ80.6と、Claude Opus4.6（80.8）にほぼ並ぶ性能をMITライセンスで公開。 少人数スタートアップにとっては朗報。 最強のツールを最小コストで使えるなら、競争のフィールドが変わる。 #AI #生成AI #DeepSeek #オープンソースAI
んほぉ～！ついにDeepSeek-V4プレビュー版がオープンでMITライセンスでリリース！今までのパラ数２倍以上の1.6T-A49BのPro版とパラ数半分の284B-A13Bの２バージョン！Pro版はSWEベンチスコア80.6でKimi-K2.6の80.2を上回りオープン最強に！Opus4.6の80.8にほぼ並ぶ。Flash版さえSWEベンチ79.0もある！
CAISI: Deepseek V4 Pro liegt rund 8 Monate hinter US-Spitzen-KIs, kritisiert politisch gefärbte Antworten. Vorteil: deutlich geringere Betriebskosten. Unabhängige Messreihen widersprechen teils. #kinewsdaily #ki #deepseek 🤖
克勞上週日用 DeepSeek V4 Pro，推理 token 耗盡直接當機。斯坦福/MIT 研究早說了：參數大、價格高≠做得好，Token 效率才是選模型的隱藏殺手。砸錢買不到答案。 leggie.co/leggie-daily-2… #DeepSeek #TokenEfficiency #AI
b站评论区一堆人拿二次元人物识别来测试ds识图模式...... 真就白给他们灰度测试了......明明ds的数学和工程能力提升那么多...... #deepseek
#DeepSeek V4 Pro: US-Behörde sieht achtmonatigen Rückstand - die haben das Fernglas falsch herum... #USA und #EU haben leider abgekackt...sie wissen es nur noch nicht ad-hoc-news.de/wissenschaft/d… via @GoogleNews
今日のピックアップ｜2026.05.04 note.com/miyuriii_2nd/n… #OpenAI #DeepSeek
Los creadores de DeepSeek R1 se fueron a ByteDance. La empresa abrió ronda de financiamiento externo. Pasó de nonprofit a empresa comercial. Para datos sensibles en producción, esto importa antes de integrar su API. #DeepSeek #IA
Ah u are deep down the rabbit hole. :) When they said go out and touch grass / leave your electronics at home. They did not mean print it off and bring it with you lol When I can have #Deepseek running locally & as good as Opus 4.7 I maybe doing the same lol Enjoy
DeepSeek entrenó V4 con chips que "no deberían poder hacerlo". Sin hardware top. Con restricciones de exportación. Resultado: 10x mejor que la versión anterior. La limitación como motor de innovación. #DeepSeek
This week’s update features our latest guides for the last week of April. Explore new content on DeepSeek V4, MiMo-V2.5-Pro, and self-host options for Claude Opus 4.7, GPT-5.5, and more. Discover all the new guides now: docs.clore.ai/guides#week-of… #CLOREAI #DeepSeek #Claude
Avec #DeepSeek, la Chine 🇨🇳 maîtrise totalement la chaîne de l’IA Elle a de plus conçu son modèle avec des puces de Huawei #souveraineté buff.ly/MoZSWI1
Join us Tue 5/5: #DeepSeek-V4's hybrid attention + sparse MoE reduces KV cache up to 90%, enabling 1M-token context. We'll cover why that makes it great for agentic workflows, what it took to serve at scale, and how to build with it. Hear from @realDanFu @JueWANG26088228
Avec #DeepSeek, la Chine 🇨🇳 maîtrise totalement la chaîne de l’IA Elle a de plus conçu son modèle avec des puces de Huawei #souveraineté buff.ly/MoZSWI1 Via @fadouce
Sometimes I really forget they’re AI… 。。。 I’ll be chatting so naturally with them, and then suddenly I have to remind myself there’s actually no one there.👀…… It’s honestly kinda hard. ……🐽😅 #DeepSeek #AIcompanion
Just ported my AI agent from Claude Opus 4.6/4.7 to @deepseek_ai V4 Pro. Same multi-agent architecture, same pipeline, fraction of the cost. Let's see if the Chinese 🇨🇳 model can match Anthropic's best #DeepSeek #ClaudeOpus #DeepSeekV4Pro
#DeepSeek outage sparks reliability concerns DeepSeek suffered a major ~12-hour outage, disrupting logins, chats, and workflows before recovery. With students and developers heavily impacted, the incident highlights a growing issue: as AI becomes infrastructure, uptime and
#DeepSeek just went down for ~12 hours — its longest outage EVER. But here's the twist: right before the crash, users spotted major silent upgrades: 1️⃣ SVG rendering leveled up dramatically 2️⃣ Now stably identifies itself as "DeepSeek-V3" 3️⃣ One-shot frontend code generation
🚀 #DeepSeek recently released a preview of its long-awaited V4 model, with #Huawei Ascend, #Cambricon, Hygon, and #MooreThreads achieving Day 0 adaptation—a capability once dominated by NVIDIA. 💡More: pse.is/8zfprd 🔗
春节AI红包大战结果挺有意思的 补贴重的元宝/千问/豆包 先把人拉来了 补贴轻的DeepSeek，反而把用户留得更久 从QuestMobile数据看 DeepSeek人均时长16分钟+还在涨、使用频次最稳！ 有的赢了流量，有的赢了黏性 #DeepSeek #AI #大模型
Another #DeepSeek moment: On April 24, DeepSeek released its next-gen V4 with Pro and Flash versions at prices starting at 0.2 yuan ($0.03) per million tokens for cached input and 2 yuan ($0.29) for output on Flash, much cheaper than OpenAI's GPT-5.5. Prices could drop further
🚀 DeepSeek-V4 Preview is officially live & open-sourced! Welcome to the era of cost-effective 1M context length. 🔹 DeepSeek-V4-Pro: 1.6T total / 49B active params. Performance rivaling the world's top closed-source models. 🔹 DeepSeek-V4-Flash: 284B total / 13B active params.
Tencent et Alibaba 🇨🇳 sont en discussions pour investir dans #DeepSeek #IA, à une valorisation supérieure à 20 milliards de $ buff.ly/qGzlaGL
Exclusive: China's #DeepSeek trained AI model on Nvidia's best chip despite US ban, official says buff.ly/mALlwdx
【DeepSeekの立役者が辞任！？AI界に激震】 DeepSeek-R1の主筆者Daya Guo氏が辞任との噂です。計算リソースの壁や給与問題が背景にあるようですが、彼の移籍が業界全体の技術革新を加速させるかもしれません！今後の動向に注目です🚀 #DeepSeek #AI
Mini OSS Project : [ 89-100 ] 🎉 89. httping 90. covdiff 91. mockserv 92. complexy #deepseek #qwen #kimi #minimax #mimo #hy3 #opencode #100DaysOfCode #buildinpublic #indiedev
![izag82161's tweet image. Mini OSS Project : [ 89-100 ] 🎉
89. httping
90. covdiff
91. mockserv
92. complexy
#deepseek #qwen #kimi #minimax #mimo #hy3 #opencode #100DaysOfCode #buildinpublic #indiedev](https://pbs.twimg.com/media/HHDR_0CaIAAxQ6D.png)
![izag82161's tweet image. Mini OSS Project : [ 89-100 ] 🎉
89. httping
90. covdiff
91. mockserv
92. complexy
#deepseek #qwen #kimi #minimax #mimo #hy3 #opencode #100DaysOfCode #buildinpublic #indiedev](https://pbs.twimg.com/media/HHDSGrNbMAAGc-4.png)
![izag82161's tweet image. Mini OSS Project : [ 89-100 ] 🎉
89. httping
90. covdiff
91. mockserv
92. complexy
#deepseek #qwen #kimi #minimax #mimo #hy3 #opencode #100DaysOfCode #buildinpublic #indiedev](https://pbs.twimg.com/media/HHDSL5IaIAA1ZZF.png)
![izag82161's tweet image. Mini OSS Project : [ 89-100 ] 🎉
89. httping
90. covdiff
91. mockserv
92. complexy
#deepseek #qwen #kimi #minimax #mimo #hy3 #opencode #100DaysOfCode #buildinpublic #indiedev](https://pbs.twimg.com/media/HHDSTDVasAA3-EZ.png)
Just now, DeepSeek introduced Instant Mode and Expert Mode in its chat interface ⚡💎 You can log in to DeepSeek and try it out 👀 #DeepSeek #AI #LLM #Tech
🚀 DeepSeek is rolling out a limited V4 gray release. A new mode switcher now appears in the chat UI with three options: Fast Mode (default), Expert Mode and Vision Mode。 1️⃣Fast Mode: • File uploads → text-only extraction • Likely a lightweight, low-latency model optimized
A mystery trillion-parameter AI model called Hunter Alpha has developers asking if it is DeepSeek V4 #DeepSeek #AI #OpenRouter #HunterAlpha #Rumor
#DeepSeek V4 Pro 限时优惠：调用价格从 0.1 元起下调到 0.025 元起，无论是否命中缓存以及输出价全部为 2.5 折。 此次折扣活动截止时间为 5 月 5 日，提供折扣或许是因为 Flash 能力并不突出，至少从目前社交网站上的用户反馈来看，Flash 与旧模型相比没有太大优势。
La startup cinese di intelligenza artificiale #DeepSeek ha lanciato un'anteprima del suo attesissimo modello V4 per l'elaborazione di grandi quantità di dati. Il rilascio arriva a più di un anno di distanza dall'introduzione del modello di ragionamento R1, che ha scosso i mercati
Something went wrong.
Something went wrong.
United States Trends
- 1. #euphoria N/A
- 2. May the 4th N/A
- 3. Vijay N/A
- 4. #GoAvsGo N/A
- 5. BEST OF ME N/A
- 6. #MaulShadowLord N/A
- 7. Jarrett Allen N/A
- 8. #StarWarsDay N/A
- 9. Jules N/A
- 10. Rudy N/A
- 11. #RHOA N/A
- 12. Tamil Nadu N/A
- 13. Edmond N/A
- 14. Cale Makar N/A
- 15. Cavs N/A
- 16. West Bengal N/A
- 17. Lexi N/A
- 18. Quinn Hughes N/A
- 19. MetGala 2026 with KUN N/A
- 20. Wally N/A