• Home
  • Technology
  • Gaming
  • Entertainment
  • World & Business
  • Science
  • Sports
  • AI
HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • HomeTechnologyGamingEntertainmentWorld & BusinessScienceSportsAI
    • Home
    • Technology
    • Gaming
    • Entertainment
    • World & Business
    • Science
    • Sports
    • AI
    AI

    DeepSeek Releases V4-Flash With Agent Upgrades

    Commentators discuss DeepSeek V4-Flash benchmarks, agent upgrades, and cost comparisons.

    GN
    JH
    AS
    112 Sources, 62d ago, first seen 62d ago

    TLDR

    DeepSeek posted that it has massively upgraded agent capabilities in V4-Flash, with benchmark scores now far surpassing V4-Pro-Preview. Zephyr reported a 76.7 score on Cybergem for a 200B model. TeortaxesTex stated the model is 14 times cheaper and 50 percent faster than V4-Pro. Arena.ai listed DeepSeek-V4-Flash-High at 1586 points, ranking seventh overall and third among open models on the Frontend Code Arena leaderboard. Bindu Reddy called it a strong small model that competes with GPT-Luna but appears benchmark-maxed. Chamath referenced a 105 times lower cost claim versus another system.

    Combined views

    13.7M

    112 Sources, first seen 62d ago

    Combined views

    13.7M

    112 Sources, first seen 62d ago

    79.6K likes
    79.6K likes
    3.7K comments
    12.5K saves
    9.6K reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    3.7K comments
    12.5K saves
    9.6K reposts
    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    112 Sources

    @zephyr_z9V4 flash update dropped 76.7 on Cybergem for a 200B model Huh????
    @teortaxesTexDeepSeek-V4-Flash is updated. «DSBench-FullStack is an internal full-stack development test set, and DSBench-Hard is an internal Coding Agent hard-problem test set» «The official release of DeepSeek-V4-Pro will follow soon» These scores are… quite a lot better than Preview.
    @sundeepView post on X
    @deepseek_ai🚀 DeepSeek-V4-Flash Official API is now LIVE in public beta! 🔷 We’ve massively upgraded its Agent capabilities—benchmark scores are now far surpassing the V4-Pro-Preview. Check out the massive performance leap below! 👇 🔷 The official V4-Flash now natively supports the Responses API format and is fully adapted for Codex! Check out the configuration details in our official API docs: https://api-docs.deepseek.com/quick_start/agent_integrations/codex
    @iScienceLuvr"We’ve massively upgraded its Agent capabilities—benchmark scores are now far surpassing the V4-Pro-Preview."
    @xeophonit is exactly 50 but for a whopping 0.20 proven wrong, sad!
    @kimmonismusThis is insane! not kidding at all. DeepSeek v4 *Flash* super close to Opus 4.8, Insane upgrade - DeepSWE 54.4%, outperforming GLM-5.2, its 4 Pro version by a lot, almost on par with Opus 4.8 - TerminalBench 82.7%, outperforming 4 Pro, GLM-5.2 and again super close to Opus 4.8 Price: in $0.28 per 1m / out $0.87 per 1m This is probably the price-performance release in reaction to OpenAIs Luna / Terra price drop INSANE.
    @nrehiew_This is likely the best post training DeepSeek has ever done
    @rohanpaul_aiDeepSeek-V4-Flash is now live as a public beta API, s 82.7 on Terminal Bench 2.1 and beating the larger Pro-Preview. the reason to route expensive work to the bigger sibling is gone, and long agent loops are where spend piles up.
    @Yampelegthis is getting ridiculous

    112 Sources

    @zephyr_z9V4 flash update dropped 76.7 on Cybergem for a 200B model Huh????
    @teortaxesTexDeepSeek-V4-Flash is updated. «DSBench-FullStack is an internal full-stack development test set, and DSBench-Hard is an internal Coding Agent hard-problem test set» «The official release of DeepSeek-V4-Pro will follow soon» These scores are… quite a lot better than Preview.
    @sundeepView post on X
    @deepseek_ai🚀 DeepSeek-V4-Flash Official API is now LIVE in public beta! 🔷 We’ve massively upgraded its Agent capabilities—benchmark scores are now far surpassing the V4-Pro-Preview. Check out the massive performance leap below! 👇 🔷 The official V4-Flash now natively supports the Responses API format and is fully adapted for Codex! Check out the configuration details in our official API docs: https://api-docs.deepseek.com/quick_start/agent_integrations/codex
    @iScienceLuvr"We’ve massively upgraded its Agent capabilities—benchmark scores are now far surpassing the V4-Pro-Preview."
    @xeophonit is exactly 50 but for a whopping 0.20 proven wrong, sad!
    @kimmonismusThis is insane! not kidding at all. DeepSeek v4 *Flash* super close to Opus 4.8, Insane upgrade - DeepSWE 54.4%, outperforming GLM-5.2, its 4 Pro version by a lot, almost on par with Opus 4.8 - TerminalBench 82.7%, outperforming 4 Pro, GLM-5.2 and again super close to Opus 4.8 Price: in $0.28 per 1m / out $0.87 per 1m This is probably the price-performance release in reaction to OpenAIs Luna / Terra price drop INSANE.
    @nrehiew_This is likely the best post training DeepSeek has ever done
    @rohanpaul_aiDeepSeek-V4-Flash is now live as a public beta API, s 82.7 on Terminal Bench 2.1 and beating the larger Pro-Preview. the reason to route expensive work to the bigger sibling is gone, and long agent loops are where spend piles up.
    @Yampelegthis is getting ridiculous