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

    Multimodal Decision model announced for open source and Decisions API access

    Arav Srinivas claims it scores higher than Jev at a fraction of the price: 4 cents per million input tokens.

    AS
    B(
    DY
    7 Sources, ,

    TLDR

    Arav Srinivas says a multimodal Decision model is being open-sourced and offered through the Decisions API. He claims it scores higher than Jev at a fraction of the price, listing 4 cents per million input tokens.

    Combined views

    88.7K

    7 Sources, first seen 3h ago

    591 likes

    Combined views

    88.7K

    7 Sources, first seen 3h ago

    591 likes
    3h ago
    first seen 3h ago
    54 comments
    250 saves
    87 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Featured Source
    54 comments
    250 saves
    87 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    #11

    Today's Rank

    #11

    7 Sources

    @perplexitydevsIntroducing the Perplexity Decisions API. It's powered by pplx-decider-v1-27b, our multimodal decision model trained to output a probability distribution over a fixed set of answers instead of text. It costs $0.04/million input tokens and scores 85.71% across benchmarks.
    @beffjezosNew open source Jev-like model!
    @krishnanrohitEven cheaper and better decisions !!
    @AravSrinivasRT @perplexitydevs: Introducing the Perplexity Decisions API. It's powered by pplx-decider-v1-27b, our multimodal decision model trained t…
    @denisyaratswe open-sourced pplx-decider-v1-27b, a powerful generative discriminator. it's trained on top of qwen3.8-27b, so it natively supports 250k context and is multimodal (e.g., it can analyze images). its performance is strong on public held-out benchmarks and on our internal benchmarks as well. one cool use case we found: we use a set of classifiers based on this model to monitor our RL rollouts, which speeds up training quite a bit and helps us fix problems with RL environments and model behavior. we are also making this model available in the Perplexity API. it's very fast and cheap ($0.04/1M tokens), and we will make it even cheaper in the next few days. here is a fun demo of chess self-play: pplx-decider-v1-27b plays both sides, sees only a screenshot of the board, and picks each move in ~140ms through our API. the video is in real time.

    7 Sources

    @perplexitydevsIntroducing the Perplexity Decisions API. It's powered by pplx-decider-v1-27b, our multimodal decision model trained to output a probability distribution over a fixed set of answers instead of text. It costs $0.04/million input tokens and scores 85.71% across benchmarks.
    @beffjezosNew open source Jev-like model!
    @krishnanrohitEven cheaper and better decisions !!
    @AravSrinivasRT @perplexitydevs: Introducing the Perplexity Decisions API. It's powered by pplx-decider-v1-27b, our multimodal decision model trained t…
    @denisyaratswe open-sourced pplx-decider-v1-27b, a powerful generative discriminator. it's trained on top of qwen3.8-27b, so it natively supports 250k context and is multimodal (e.g., it can analyze images). its performance is strong on public held-out benchmarks and on our internal benchmarks as well. one cool use case we found: we use a set of classifiers based on this model to monitor our RL rollouts, which speeds up training quite a bit and helps us fix problems with RL environments and model behavior. we are also making this model available in the Perplexity API. it's very fast and cheap ($0.04/1M tokens), and we will make it even cheaper in the next few days. here is a fun demo of chess self-play: pplx-decider-v1-27b plays both sides, sees only a screenshot of the board, and picks each move in ~140ms through our API. the video is in real time.