ru
Feedback
Web 3.0 Ethiopia - DeFi & AI

Web 3.0 Ethiopia - DeFi & AI

Открыть в Telegram

Bridging the Information Gap on DeFI and Artificial Intelligence for Ethiopians

Больше
699
Подписчики
-124 часа
Нет данных7 дней
-230 день
Архив постов
Anthropic Extends Free Promotional Access for Claude Fable 5 Until July 12 Anthropic has announced an extension of its promot
Anthropic Extends Free Promotional Access for Claude Fable 5 Until July 12 Anthropic has announced an extension of its promotional access period for Claude Fable 5, its newest and most advanced model. All paid subscribers can now continue to use the highly capable model at no extra cost through July 12, 2026, at 11:59 PM PT. During this promotional window, paid users can allocate up to 50% of their standard weekly subscription limits exclusively to Fable 5. Once this limit is reached, users have the flexibility to either continue using the model by purchasing additional usage credits or simply switch back to other models within their plan. አንትሮፒክ አዲሱን እና እጅግ ዘመናዊ የሆነውን "ክሎድ ፌብል 5" ሞዴል ያለ ተጨማሪ ክፍያ የመሞከሪያ ጊዜን ማራዘሙን አስታወቀ። ማንኛውም ክፍያ የሚፈጽም ተጠቃሚ እስከ ጁላይ 12 ቀን 2026፣ ከምሽቱ 5፡59 ሰዓት ድረስ ይህን የላቀ ሞዴል ያለ ተጨማሪ ክፍያ መጠቀም ይችላል። በዚህ የፕሮሞሽን ጊዜ ውስጥ፣ ተጠቃሚዎች በሳምንት ከሚፈቀድላቸው የአጠቃቀም መጠን ውስጥ እስከ 50 በመቶ የሚሆነውን ለ"ፌብል 5" ማዋል ይችላሉ። ይህንን የነጻ ገደብ ከጨረሱ በኋላ፣ ተጨማሪ የክፍያ ክሬዲቶችን በመግዛት መጠቀምን መቀጠል ወይም ወደ ሌሎች የክሎድ ሞዴሎች መመለስ ይችላሉ። @webthreeth

Google Upgrades Managed Agents in Gemini API for Production-Ready Autonomous Workflows Google has announced significant upgrades to its Managed Agents in the Gemini API, focusing on making autonomous AI workflows more reliable and production-ready. Powered by the Antigravity engine, the update introduces asynchronous background execution for long-running tasks, allowing applications to poll for status without keeping HTTP connections open. The API now seamlessly integrates with remote Model Context Protocol (MCP) servers, enabling agents to securely access private databases and internal tools directly from isolated cloud sandboxes. ጎግል (Google) ራሳቸውን የቻሉ የኤአይ ወኪሎች (autonomous AI workflows) ይበልጥ አስተማማኝ እና ለስራ ዝግጁ እንዲሆኑ ለማድረግ በጄሚኒ ኤፒአይ (Gemini API) ውስጥ የሚገኘውን የ"ማኔጅድ ኤጀንት" (Managed Agents) አገልግሎት ማሻሻሉን አስታውቋል። በ"አንቲግራቪቲ" (Antigravity) ሞተር የሚመራው ይህ ማሻሻያ፣ ረጅም ጊዜ የሚወስዱ ተግባራትን ከበስተጀርባ (background) ለማከናወን የሚያስችል አሰራርን አካቷል፤ ይህም መተግበሪያዎች የኢንተርኔት መስመር (HTTP connections) ክፍት አድርገው ሳይጠብቁ የስራውን ሂደት እንዲከታተሉ ያስችላቸዋል። አዲሱ ኤፒአይ አሁን ላይ ከ'ኤምሲፒ' (MCP) ሰርቨሮች ጋር በቀጥታ በመገናኘት፣ የኤአይ ወኪሎች ደህንነቱ በተጠበቀ የክላውድ መድረክ ላይ ሆነው የግል የመረጃ ቋቶችን (private databases) እና የውስጥ መተግበሪያዎችን እንዲያገኙ ያስችላቸዋል። @webthreeth

First news in seven days 😅sorry for the delay. anyways...

Meta's Next Frontier Model 'Watermelon' Matches GPT-5.5 Capabilities Meta's highly anticipated next frontier model, internall
Meta's Next Frontier Model 'Watermelon' Matches GPT-5.5 Capabilities Meta's highly anticipated next frontier model, internally codenamed "Watermelon" and associated with the Llama 5 family, is currently in training using approximately 10 times the computing power of its predecessor, Muse Spark (Avocado). According to leaked statements from Chief AI Officer Alexandr Wang, the new model already matches the performance of OpenAI's GPT-5.5 on key internal benchmarks. The leaks suggest massive improvements in agentic workflows and coding capabilities. የሜታ (Meta) ቀጣይ ግዙፍ የኤአይ (AI) ሞዴል፣ በውስጥ መጠሪያው "ዎተርሜሎን" (Watermelon) የሚባለው እና ከ"ላማ 5" (Llama 5) ቤተሰብ ጋር የተቆራኘው ሞዴል፣ ከቀድሞው 'ሚዩዝ ስፓርክ' (Muse Spark) በ10 እጥፍ በሚበልጥ የኮምፒውተር አቅም እየሰለጠነ ይገኛል። የሜታ የኤአይ ዋና ሀላፊ (Chief AI Officer) አሌክሳንደር ዋንግ እንዳሉት፣ ይህ አዲስ ሞዴል በውስጥ የሙከራ መለኪያዎች ከኦፕንኤአይ "GPT-5.5" ጋር የሚስተካከል አቅም አሳይቷል። ሾልከው የወጡ መረጃዎች እንደሚያመለክቱት፣ ሞዴሉ በኤጀንት ስራዎች እና በኮዲንግ ላይ ከፍተኛ ማሻሻያዎችን ያመጣል። @webthreeth

sol sol sol sol sol sol 😅

An exciting week is brewing ahead.

Anthropic Redeploys Claude Fable 5 Globally Following US Export Control Suspension Anthropic is officially redeploying Claude
Anthropic Redeploys Claude Fable 5 Globally Following US Export Control Suspension Anthropic is officially redeploying Claude Fable 5 globally on July 2, 2026, after the model was temporarily suspended on June 12 due to US export controls concerning cybersecurity risks. Following extensive discussions with government officials, the model has been updated with advanced classifiers designed to aggressively block potential misuse cases. During this initial rollout, complex coding and debugging tasks will be temporarily routed to the Opus 4.8 model while the new safeguards are fine-tuned. አንትሮፒክ (Anthropic) በሳይበር ደህንነት ስጋት ምክንያት በጁን 12 በአሜሪካ የኤክስፖርት ህግ ታግዶ የነበረውን "ክሎድ ፌብል 5" (Claude Fable 5) ጁላይ 2 ቀን 2026 በዓለም አቀፍ ደረጃ በድጋሚ አገልግሎት ላይ አውሏል። ከመንግስት ባለስልጣናት ጋር ከተደረጉ ውይይቶች በኋላ፣ ቴክኖሎጂውን አላግባብ መጠቀምን ለመከላከል የሚያስችሉ አዳዲስ የቁጥጥር ስልቶች (classifiers) በሞዴሉ ውስጥ ተካትተዋል። አዲሱ የደህንነት ስርዓት ትክክለኛነቱ እስኪረጋገጥ ድረስ፣ የኮዲንግ እና የሶፍትዌር ማረም ስራዎች ለጊዜው ወደ "ኦፐስ 4.8" (Opus 4.8) ሞዴል እንዲመሩ ይደረጋል። @webthreeth

Anthropic Launches Claude Science to Revolutionize AI-Powered Research Workflows Anthropic has announced the beta launch of C
Anthropic Launches Claude Science to Revolutionize AI-Powered Research Workflows Anthropic has announced the beta launch of Claude Science, a dedicated application designed specifically to empower researchers and manage comprehensive scientific workflows. The new platform enables users to generate traceable artifacts that are directly linked to underlying code, ensuring absolute transparency and reproducibility. አንትሮፒክ የሳይንሳዊ ምርምር ስራዎችን ለማቀላጠፍ እና የተሟላ የስራ ፍሰትን ለማስተዳደር የተሰራ "ክሎድ ሳይንስ" (Claude Science) የተሰኘ አዲስ መተግበሪያን በሙከራ ደረጃ (beta) ይፋ አድርጓል። ይህ መድረክ ተመራማሪዎች የሚያገኟቸውን ውጤቶች እና መረጃዎች በቀጥታ ከኮድ ጋር በማስተሳሰር (traceable artifacts)፣ የምርምራቸውን ትክክለኛነት እና ግልጽነት ለማረጋገጥ ያስችላቸዋል። @webthreeth

Anthropic has announced the beta launch of Claude Science, a dedicated application designed specifically to empower researchers and manage comprehensive scientific workflows. The new platform enables users to generate traceable artifacts that are directly linked to underlying code, ensuring absolute transparency and reproducibility. Featuring on-demand compute environments and seamless connections to over 60 scientific databases, Claude Science equips researchers with an advanced, end-to-end AI workspace tailored for rigorous data analysis and academic discovery.

Google Launches Nano Banana 2 Lite and Gemini Omni Flash Google's Gemini has officially introduced Nano Banana 2 Lite and Gem
Google Launches Nano Banana 2 Lite and Gemini Omni Flash Google's Gemini has officially introduced Nano Banana 2 Lite and Gemini Omni Flash, making advanced generative media models available to developers via the Gemini API and AI Studio. Nano Banana 2 Lite is built for high-speed, cost-effective image generation, producing visuals in under 4 seconds. Meanwhile, Gemini Omni Flash provides state-of-the-art video editing capabilities priced at $0.10 per second, matching the cost efficiency of Veo 3.1 Fast and marking a significant expansion in Google's versatile AI media toolkit. ጎግል"ናኖ ባናና 2 ላይት" እና "ጄሚኒ ኦምኒ ፍላሽ" የተሰኙ አዳዲስ ሞዴሎችን በ"Gemini API" እና በ"AI Studio" በኩል ለአልሚዎች ክፍት ማድረጉን አስታወቀ። ናኖ ባናና 2 ላይት እጅግ ፈጣን እና ወጪ ቆጣቢ ለሆኑ የምስል ፈጠራዎች የተነደፈ ሲሆን፣ ከ4 ሰከንድ ባነሰ ጊዜ ውስጥ ምስልን የመፍጠር እና ለ1,000 ምስሎች 0.034 ዶላር ብቻ የሚያስከፍል ነው፤ በመለኪያ ፈተናዎችም ከሌሎች ሞዴሎች የላቀ ውጤት አሳይቷል። በሌላ በኩል ጄሚኒ ኦምኒ ፍላሽ የላቀ የቪዲዮ አርትኦት አቅም ያለው ሲሆን፣ በአንድ ሰከንድ 0.10 ዶላር ብቻ በማስከፈል ጎግል በኤአይ (AI) የሚዲያ ዘርፍ ያለውን ተደራሽነት ይበልጥ እያሰፋው ይገኛል። @webthreeth

Meituan Launches LongCat-2.0 Food delivery giant Meituan has launched LongCat-2.0, a 1.6T parameter MoE coding model pre-trai
Meituan Launches LongCat-2.0 Food delivery giant Meituan has launched LongCat-2.0, a 1.6T parameter MoE coding model pre-trained entirely on over 50,000 domestic Huawei ASICs. Featuring a 1M context window and near-frontier agentic capabilities, it scores on 59.5 on SWE-bench Pro, notably edging out GPT-5.5. This landmark achievement marks the first time a frontier-class model has been successfully pre-trained on alternative hardware across 35 trillion tokens without a single training rollback. የምግብ ማድረሻው ግዙፍ ኩባንያ ሜይቱዋን ሙሉ በሙሉ ከ50,000 በላይ በሆኑ የሀገር ውስጥ የሁዋዌ ኤሲኮች ላይ አስቀድሞ የሰለጠነውንና 1.6 ትሪሊዮን መለኪያዎች ያሉትን ሎንግካት-2.0 (LongCat-2.0) የተሰኘ የኤምኦኢ (MoE) የኮዲንግ ሞዴል ይፋ አድርጓል። ባለ 1 ሚሊዮን ቶክን የዐውደ-ጽሑፍ መስኮት እና ወደ ፊት-ቀደምት ደረጃ የተቃረበ ወኪል-ተኮር (agentic) አቅም ያለው ይህ ሞዴል፣ በተርሚናል-ቤንች 2.1 ላይ 70.8 እንዲሁም በኤስደብሊውኢ-ቤንች ፕሮ ላይ 59.5 ያስመዘገበ ሲሆን፣ በተለይም ጂፒቲ-5.5ን (GPT-5.5) በበላይነት ማሸነፍ ችሏል። ይህ ታሪካዊ ስኬት የፊት-ቀደምት ደረጃ ሞዴል በሌላ አማራጭ ሃርድዌር ላይ በ35 ትሪሊዮን ቶክኖች ላይ አንድም የስልጠና መቋረጥ ወይም ወደኋላ መመለስ ሳይገጥመው በስኬት አስቀድሞ ሲሰለጥን የመጀመሪያው ነው። @webthreeth

OpenClaw Launches Privacy-First Mobile Apps for Self-Hosted AI Agents OpenClaw has announced the release of native iOS and An
OpenClaw Launches Privacy-First Mobile Apps for Self-Hosted AI Agents OpenClaw has announced the release of native iOS and Android applications that function as companion nodes, connecting users smartphones to their self-hosted AI agent gateways via WebSocket. Designed with strict privacy in mind, the apps allow users to access agents for managing emails, calendars, and automation on the go while keeping all core processing safely on their personal machines. ኦፕን ክሎው (OpenClaw) ለአይኦኤስ (iOS) እና ለአንድሮይድ (Android) ስልኮች የሚሆኑ አዳዲስ መተግበሪያዎችን ይፋ አድርጓል። እነዚህ መተግበሪያዎች የተጠቃሚዎችን ሞባይል ስልክ በግል ኮምፒውተራቸው ላይ ከሚገኘው የሰው ሰራሽ አስተውሎት ወኪል (AI agent) ጋር በቀጥታ በማገናኘት፣ ኢሜይሎችን፣ የቀን መቁጠሪያዎችን እና የዕለት ተዕለት ተግባራትን ከየትኛውም ቦታ ሆነው በስልክ ለማስተዳደር ያስችላሉ። 🍎 - https://apps.apple.com/us/app/openclaw-ai-that-does-things/id6780396132 Android - https://play.google.com/store/apps/details?id=ai.openclaw.app @webthreeth

Anthropic Upgrades Claude Code with Background Subagents for Seamless Multitasking Anthropic’s Boris Cherny has announced a m
Anthropic Upgrades Claude Code with Background Subagents for Seamless Multitasking Anthropic’s Boris Cherny has announced a major upgrade for the upcoming version of Claude Code: subagents will now run in the background by default. This enhancement allows users to continue chatting and interacting with the main Claude instance without interruption while specialized agents complete lengthy tasks asynchronously. የአንትሮፒክ ባለሙያ የሆኑት ቦሪስ ቼርኒ (Boris Cherny) በሚቀጥለው የ"ክሎድ ኮድ" (Claude Code) ስሪት ላይ፣ ንዑስ የኤአይ ወኪሎች (subagents) በነባሪነት ከበስተጀርባ (background) ሆነው ስራቸውን እንደሚያከናውኑ አስታውቀዋል። ይህ ማሻሻያ ተጠቃሚዎች አንድ ወኪል ረጅም ጊዜ የሚወስድ ስራን እያከናወነ ባለበት ሰዓት ሳይስተጓጎሉ ከዋናው ክሎድ ጋር ውይይታቸውን እና ሌሎች ተግባራትን እንዲቀጥሉ ያስችላቸዋል። @webthreeth

Elon Musk Announces Grok 4.5 in Private Beta, Aiming to Surpass Claude Opus Elon Musk has announced that Grok 4.5 is official
Elon Musk Announces Grok 4.5 in Private Beta, Aiming to Surpass Claude Opus Elon Musk has announced that Grok 4.5 is officially in private beta testing within SpaceX and Tesla. Built on a massive 1.5-trillion parameter V9 foundation model and supplemented with specialized data from Cursor, early internal evaluations suggest the new model's performance rivals or may even exceed Anthropic's top-tier Claude Opus. ኢሎን ማስክ "ግሮክ 4.5" (Grok 4.5) የተሰኘው አዲስ የሰው ሰራሽ አስተውሎት ሞዴል በስፔስኤክስ (SpaceX) እና በቴስላ (Tesla) ተቋማት ውስጥ በውስጥ የሙከራ ደረጃ (private beta) ላይ እንደሚገኝ አስታውቀዋል። በ1.5 ትሪሊዮን ፓራሜትር አቅም ባለው "V9" መሠረታዊ ሞዴል ላይ የተገነባው እና ከ"ከርሰር" (Cursor) በተገኘ ተጨማሪ መረጃ የዳበረው ይህ ሞዴል፣ ከወዲሁ የአንትሮፒክን "ክሎድ ኦፐስ" (Claude Opus) የሚስተካከል ወይም የሚበልጥ ብቃት እንዳለው የውስጥ ግምገማዎች ያሳያሉ። @webthreeth

5.6 👀
5.6 👀

Going to LinkedIn now feels like going in Moltbook.

Ethiopian AI Platform ScholarXIV Launches MCP to Connect AI Agents with Deep Research ScholarXIV, an Ethiopian AI platform, h
Ethiopian AI Platform ScholarXIV Launches MCP to Connect AI Agents with Deep Research ScholarXIV, an Ethiopian AI platform, has officially launched its highly anticipated ScholarXIV MCP (Model Context Protocol), providing AI agents with direct access to a comprehensive suite of academic research tools. Featuring over 30 built-in tools, the MCP allows agents to search papers, manage collections, read metadata, and even execute deep research workflows autonomously. "ስኮላር 14" (ScholarXIV) የተሰኘው የኢትዮጵያ የሰው ሰራሽ አስተውሎት (AI) መድረክ፣ የኤአይ ወኪሎች በቀጥታ የአካዳሚክ ምርምር መረጃዎችን እንዲያገኙ የሚያስችለውን 'ኤምሲፒ' (Model Context Protocol - MCP) በይፋ አስጀመረ። ከ30 በላይ መሳሪያዎችን (tools) ያካተተው ይህ ሲስተም፣ የኤአይ ወኪሎች የጥናት ወረቀቶችን እንዲፈልጉ፣ መረጃዎችን እንዲያደራጁ እና ጥልቅ የምርምር ስራዎችን በራስ-ሰር እንዲያከናውኑ ያስችላቸዋል። @webthreeth

Google Enhances Gemini 3.5 Flash With Native Computer Use Capabilities for Seamless Cross-Platform Enterprise Automation Gemi
Google Enhances Gemini 3.5 Flash With Native Computer Use Capabilities for Seamless Cross-Platform Enterprise Automation Gemini 3.5 Flash now natively integrates computer use capabilities, allowing custom agents to see, reason, and execute tasks across browser, mobile, and desktop environments. This upgrade from the previous standalone model delivers frontier-level accuracy with superior speed and cost-efficiency for long-horizon enterprise automation like software testing. ጀሚናይ 3.5 ፍላሽ (Gemini 3.5 Flash) ብጁ የሆኑ ወኪሎች (custom agents) በብሮውዘር፣ በሞባይል እና በዴስክቶፕ አካባቢዎች ላይ ማየት፣ ማመዛዘን እና ተግባራትን መፈጸም እንዲችሉ የሚያደርገውን የኮምፒውተር አጠቃቀም ብቃትን (computer use capabilities) አሁን በተፈጥሯዊ ሁኔታ አዋህዶ ይዟል። ይህ ከቀደመው ራሱን ከቻለ ሞዴል የተደረገው ማሻሻያ እንደ ሶፍትዌር ሙከራ (software testing) ላሉ ረጅም ጊዜ ለሚወስዱ የድርጅት አውቶሜሽን ስራዎች የፊት-ቀደምት ደረጃ ትክክለኛነትን ከላቀ ፍጥነት እና ወጪ ቆጣቢነት ጋር ያቀርባል። @webthreeth

OpenAI Unveils Jalapeño Custom AI Chip to Cut Operating Costs by Half OpenAI has officially unveiled Jalapeño, its first in-h
OpenAI Unveils Jalapeño Custom AI Chip to Cut Operating Costs by Half OpenAI has officially unveiled Jalapeño, its first in-house Intelligence Processor developed in collaboration with semiconductor giant Broadcom and systems manufacturer Celestica. Designed specifically for Large Language Model inference, this custom chip was built from scratch in just nine months with the help of OpenAI's own AI models. Early lab tests indicate the hardware can cut AI operating costs by roughly 50 percent compared to standard GPUs currently on the market while maximizing performance per watt. ኦፕንኤአይ (OpenAI) ከብሮድኮም (Broadcom) እና ከሰሌስቲካ (Celestica) ጋር በመተባበር ያበለጸገውን "ጃላፔኖ" (Jalapeño) የተሰኘ የመጀመሪያውን የውስጥ የሰው ሰራሽ አስተውሎት ፕሮሰሰር (chip) በይፋ አስተዋውቋል። በተለይ ለግዙፍ የቋንቋ ሞዴሎች (LLMs) ትንተና የተነደፈው ይህ ቺፕ በኩባንያው የራሱ የኤአይ ሞዴሎች እገዛ በዘጠኝ ወራት ውስጥ ብቻ ከባዶ የተገነባ ነው። የመጀመሪያ የላቦራቶሪ ሙከራዎች እንደሚያሳዩት ይህ አዲስ ቴክኖሎጂ አሁን በገበያ ላይ ካሉ መደበኛ ጂፒዩዎች (GPUs) ጋር ሲነጻጸር የኤአይ አሰራር ወጪን በ50 በመቶ ሊቀንስ ይችላል። @webthreeth

Anthropic Launches Claude Tag to Bring AI Agents Directly into Slack Anthropic has introduced Claude Tag for its Enterprise a
Anthropic Launches Claude Tag to Bring AI Agents Directly into Slack Anthropic has introduced Claude Tag for its Enterprise and Team users, seamlessly integrating the AI as a taggable team member directly within Slack. By simply mentioning @Claude in any channel, users can delegate complex tasks like writing code, data analysis, and incident resolution. The feature breaks requests into manageable stages, builds persistent context for smooth team handoffs, and even offers an ambient mode for proactive follow ups on quiet discussions. አንትሮፒክ ለኤንተርፕራይዝ (Enterprise) እና ለቲም (Team) ተጠቃሚዎቹ ኤአይን እንደ የስራ ባልደረባ በቀጥታ በስላክ (Slack) ውስጥ የሚያካትተውን ክሎድ ታግ (Claude Tag) ይፋ አድርጓል። ተጠቃሚዎች በማንኛውም የውይይት መስኮት ውስጥ ስሙን በመጥቀስ ብቻ እንደ ኮዲንግ መጻፍ፣ መረጃ መተንተን እና የቴክኒክ ችግሮችን መፍታት ያሉ ውስብስብ ስራዎችን እንዲሰራ ማዘዝ ይችላሉ። ይህ አገልግሎት ትዕዛዞችን በደረጃ በመከፋፈል ያከናውናል፣ ለሌሎች የቡድን አባላት መረጃዎችን በተደራጀ መልኩ ያስተላልፋል፣ እንዲሁም ጸጥ ባሉ ውይይቶች ላይ በራሱ ጣልቃ ገብቶ ማሳሰቢያዎችን የሚሰጥበት አማራጭ አለው። @webthreeth