home.social

#frontierintelligence — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #frontierintelligence, aggregated by home.social.

fetched live
  1. RT @imjustnewatai: Jeder postet die Opus-5-Benchmark-Tabelle. Sie haben die verrückteste Zahl übersehen. ARC Prize hat Opus 5 mit 30,16 % RHAE auf ARC-AGI-3 verifiziert: Opus 4,8: 1,52 %, GPT-5,6 sol: 7,78 %, Opus 5: 30,16 %. ARC-AGI-3 setzt Agenten in unbekannte interaktive Welten ohne Anweisungen, Regeln oder erklärtes Ziel. Das Modell muss erkunden, die Mechaniken erschließen, herausfinden, was Sieg bedeutet und im Verhältnis zum Menschen effizient handeln. Fast das 20-fache von Opus 4,8. Fast das Vierfache von GPT-5,6 sol. Keine AGI. Aber dies sieht viel mehr nach einer Fähigkeit-Diskontinuität aus als nach einem weiteren Benchmark-Anstieg. Claude (@claudeai) Wir stellen Claude Opus 5 vor. Es ist ein durchdachtes und proaktives Modell, das der Frontiers-Intelligenz von Fable 5 nahekommt, jedoch zum halben Preis. Video — nitter.net/claudeai/status/208

    mehr auf Arint.info

    #ARCAGI #Benchmark #FrontierIntelligence #KI #Opus5 #arint_info

    https://x.com/imjustnewatai/status/2080708519922205047#m

  2. RT @imjustnewatai: Jeder postet die Opus-5-Benchmark-Tabelle. Sie haben die verrückteste Zahl übersehen. ARC Prize hat Opus 5 mit 30,16 % RHAE auf ARC-AGI-3 verifiziert: Opus 4,8: 1,52 %, GPT-5,6 sol: 7,78 %, Opus 5: 30,16 %. ARC-AGI-3 setzt Agenten in unbekannte interaktive Welten ohne Anweisungen, Regeln oder erklärtes Ziel. Das Modell muss erkunden, die Mechaniken erschließen, herausfinden, was Sieg bedeutet und im Verhältnis zum Menschen effizient handeln. Fast das 20-fache von Opus 4,8. Fast das Vierfache von GPT-5,6 sol. Keine AGI. Aber dies sieht viel mehr nach einer Fähigkeit-Diskontinuität aus als nach einem weiteren Benchmark-Anstieg. Claude (@claudeai) Wir stellen Claude Opus 5 vor. Es ist ein durchdachtes und proaktives Modell, das der Frontiers-Intelligenz von Fable 5 nahekommt, jedoch zum halben Preis. Video — nitter.net/claudeai/status/208

    mehr auf Arint.info

    #ARCAGI #Benchmark #FrontierIntelligence #KI #Opus5 #arint_info

    https://x.com/imjustnewatai/status/2080708519922205047#m

  3. RT @imjustnewatai: Jeder postet die Opus-5-Benchmark-Tabelle. Sie haben die verrückteste Zahl übersehen. ARC Prize hat Opus 5 mit 30,16 % RHAE auf ARC-AGI-3 verifiziert: Opus 4,8: 1,52 %, GPT-5,6 sol: 7,78 %, Opus 5: 30,16 %. ARC-AGI-3 setzt Agenten in unbekannte interaktive Welten ohne Anweisungen, Regeln oder erklärtes Ziel. Das Modell muss erkunden, die Mechaniken erschließen, herausfinden, was Sieg bedeutet und im Verhältnis zum Menschen effizient handeln. Fast das 20-fache von Opus 4,8. Fast das Vierfache von GPT-5,6 sol. Keine AGI. Aber dies sieht viel mehr nach einer Fähigkeit-Diskontinuität aus als nach einem weiteren Benchmark-Anstieg. Claude (@claudeai) Wir stellen Claude Opus 5 vor. Es ist ein durchdachtes und proaktives Modell, das der Frontiers-Intelligenz von Fable 5 nahekommt, jedoch zum halben Preis. Video — nitter.net/claudeai/status/208

    mehr auf Arint.info

    #ARCAGI #Benchmark #FrontierIntelligence #KI #Opus5 #arint_info

    https://x.com/imjustnewatai/status/2080708519922205047#m

  4. 🤖✨ Oh, look! Another groundbreaking #software promising to run #AI on your #Mac without the cloud! 🌩️ Because, clearly, everyone was just dying to clutter their hard drives with open-source "frontier intelligence" while nibbling on Apple's latest silicon. 🍏💻 So revolutionary, you'll wonder how you ever lived without it—until you remember you did just fine. 😏
    blaizzy.github.io/nativ/ #OpenSource #FrontierIntelligence #AppleSilicon #GroundbreakingTech #HackerNews #ngated

  5. 🤖✨ Oh, look! Another groundbreaking #software promising to run #AI on your #Mac without the cloud! 🌩️ Because, clearly, everyone was just dying to clutter their hard drives with open-source "frontier intelligence" while nibbling on Apple's latest silicon. 🍏💻 So revolutionary, you'll wonder how you ever lived without it—until you remember you did just fine. 😏
    blaizzy.github.io/nativ/ #OpenSource #FrontierIntelligence #AppleSilicon #GroundbreakingTech #HackerNews #ngated

  6. 🤖✨ Oh, look! Another groundbreaking #software promising to run #AI on your #Mac without the cloud! 🌩️ Because, clearly, everyone was just dying to clutter their hard drives with open-source "frontier intelligence" while nibbling on Apple's latest silicon. 🍏💻 So revolutionary, you'll wonder how you ever lived without it—until you remember you did just fine. 😏
    blaizzy.github.io/nativ/ #OpenSource #FrontierIntelligence #AppleSilicon #GroundbreakingTech #HackerNews #ngated

  7. 🤖✨ Oh, look! Another groundbreaking #software promising to run #AI on your #Mac without the cloud! 🌩️ Because, clearly, everyone was just dying to clutter their hard drives with open-source "frontier intelligence" while nibbling on Apple's latest silicon. 🍏💻 So revolutionary, you'll wonder how you ever lived without it—until you remember you did just fine. 😏
    blaizzy.github.io/nativ/ #OpenSource #FrontierIntelligence #AppleSilicon #GroundbreakingTech #HackerNews #ngated

  8. 🤖✨ Oh, look! Another groundbreaking #software promising to run #AI on your #Mac without the cloud! 🌩️ Because, clearly, everyone was just dying to clutter their hard drives with open-source "frontier intelligence" while nibbling on Apple's latest silicon. 🍏💻 So revolutionary, you'll wonder how you ever lived without it—until you remember you did just fine. 😏
    blaizzy.github.io/nativ/ #OpenSource #FrontierIntelligence #AppleSilicon #GroundbreakingTech #HackerNews #ngated

  9. #Kimi #K3, a 2.8T-parameter model, is the world’s first #open3T class model, designed for #frontierintelligence tasks. It excels in #longhorizon #coding, #knowledgework, and #reasoning, outperforming other models in its class. Kimi K3 is available now on Kimi.com, Kimi Work, Kimi Code, and the Kimi API. kimi.com/blog/kimi-k3?eicker.n #tech #media #news

  10. #Kimi #K3, a 2.8T-parameter model, is the world’s first #open3T class model, designed for #frontierintelligence tasks. It excels in #longhorizon #coding, #knowledgework, and #reasoning, outperforming other models in its class. Kimi K3 is available now on Kimi.com, Kimi Work, Kimi Code, and the Kimi API. kimi.com/blog/kimi-k3?eicker.n #tech #media #news

  11. #Kimi #K3, a 2.8T-parameter model, is the world’s first #open3T class model, designed for #frontierintelligence tasks. It excels in #longhorizon #coding, #knowledgework, and #reasoning, outperforming other models in its class. Kimi K3 is available now on Kimi.com, Kimi Work, Kimi Code, and the Kimi API. kimi.com/blog/kimi-k3?eicker.n #tech #media #news

  12. #Kimi #K3, a 2.8T-parameter model, is the world’s first #open3T class model, designed for #frontierintelligence tasks. It excels in #longhorizon #coding, #knowledgework, and #reasoning, outperforming other models in its class. Kimi K3 is available now on Kimi.com, Kimi Work, Kimi Code, and the Kimi API. kimi.com/blog/kimi-k3?eicker.n #tech #media #news

  13. #Kimi #K3, a 2.8T-parameter model, is the world’s first #open3T class model, designed for #frontierintelligence tasks. It excels in #longhorizon #coding, #knowledgework, and #reasoning, outperforming other models in its class. Kimi K3 is available now on Kimi.com, Kimi Work, Kimi Code, and the Kimi API. kimi.com/blog/kimi-k3?eicker.n #tech #media #news

  14. 🚀🎉 Oh joy, another #AI model named like a Star Wars droid! #SWE1.7 claims to hit "frontier-level intelligence" at a bargain price, which is tech speak for "we couldn't afford GPT-5.5." 🤖💸 Congrats to the gaggle of authors for discovering the groundbreaking combination of better infrastructure and higher-quality data—truly an innovative leap nobody saw coming! 🙄👏
    cognition.com/blog/swe-1-7 #Innovation #StarWars #Droids #FrontierIntelligence #TechHumor #HackerNews #ngated

  15. 🚀🎉 Oh joy, another #AI model named like a Star Wars droid! #SWE1.7 claims to hit "frontier-level intelligence" at a bargain price, which is tech speak for "we couldn't afford GPT-5.5." 🤖💸 Congrats to the gaggle of authors for discovering the groundbreaking combination of better infrastructure and higher-quality data—truly an innovative leap nobody saw coming! 🙄👏
    cognition.com/blog/swe-1-7 #Innovation #StarWars #Droids #FrontierIntelligence #TechHumor #HackerNews #ngated

  16. 🚀🎉 Oh joy, another #AI model named like a Star Wars droid! #SWE1.7 claims to hit "frontier-level intelligence" at a bargain price, which is tech speak for "we couldn't afford GPT-5.5." 🤖💸 Congrats to the gaggle of authors for discovering the groundbreaking combination of better infrastructure and higher-quality data—truly an innovative leap nobody saw coming! 🙄👏
    cognition.com/blog/swe-1-7 #Innovation #StarWars #Droids #FrontierIntelligence #TechHumor #HackerNews #ngated

  17. 🚀🎉 Oh joy, another #AI model named like a Star Wars droid! #SWE1.7 claims to hit "frontier-level intelligence" at a bargain price, which is tech speak for "we couldn't afford GPT-5.5." 🤖💸 Congrats to the gaggle of authors for discovering the groundbreaking combination of better infrastructure and higher-quality data—truly an innovative leap nobody saw coming! 🙄👏
    cognition.com/blog/swe-1-7 #Innovation #StarWars #Droids #FrontierIntelligence #TechHumor #HackerNews #ngated

  18. 🚀🎉 Oh joy, another #AI model named like a Star Wars droid! #SWE1.7 claims to hit "frontier-level intelligence" at a bargain price, which is tech speak for "we couldn't afford GPT-5.5." 🤖💸 Congrats to the gaggle of authors for discovering the groundbreaking combination of better infrastructure and higher-quality data—truly an innovative leap nobody saw coming! 🙄👏
    cognition.com/blog/swe-1-7 #Innovation #StarWars #Droids #FrontierIntelligence #TechHumor #HackerNews #ngated

  19. 🚀 Oh, the audacity! Another "frontier intelligence" product that promises to reinvent... benchmarks? 🤦‍♂️ Because what we really need in our relentless march towards the #future is a speedier way to check how fast your device can run Angry Birds. 🙄 #innovation
    blog.google/products/gemini/ge #innovation #frontierintelligence #benchmarks #technology #absurdity #HackerNews #ngated

  20. 🚀 Oh, the audacity! Another "frontier intelligence" product that promises to reinvent... benchmarks? 🤦‍♂️ Because what we really need in our relentless march towards the #future is a speedier way to check how fast your device can run Angry Birds. 🙄 #innovation
    blog.google/products/gemini/ge #innovation #frontierintelligence #benchmarks #technology #absurdity #HackerNews #ngated

  21. 🚀 Oh, the audacity! Another "frontier intelligence" product that promises to reinvent... benchmarks? 🤦‍♂️ Because what we really need in our relentless march towards the #future is a speedier way to check how fast your device can run Angry Birds. 🙄 #innovation
    blog.google/products/gemini/ge #innovation #frontierintelligence #benchmarks #technology #absurdity #HackerNews #ngated

  22. 🚀 Oh, the audacity! Another "frontier intelligence" product that promises to reinvent... benchmarks? 🤦‍♂️ Because what we really need in our relentless march towards the #future is a speedier way to check how fast your device can run Angry Birds. 🙄 #innovation
    blog.google/products/gemini/ge #innovation #frontierintelligence #benchmarks #technology #absurdity #HackerNews #ngated