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

    Open-weight AI reportedly trails the frontier by about four months

    A post summarizing Mozilla's 91-page report says eight of OpenRouter's 10 most-used models by August token volume were open-weight. It also cites open-model use by 79% of surveyed developers.

    Rohan PaulRP
    Jonathan Richard SchwarzJR
    17 Sources, ,

    TLDR

    A September 16, 2026 post summarizing Mozilla's report says open-weight AI is about four months behind the frontier, but its usage and revenue shares differ sharply.

    The post cites roughly 20% of measured OpenRouter usage for open models versus about 4% of model-layer revenue. It says Mozilla attributes much of that mismatch to pricing, with closed models costing roughly six times more per call at about 90% capability parity. Crucially, it also relays Mozilla's warning that the revenue measurement comes from a 2025 window and predates its 2026 usage data, so the revenue split may be different.

    The post also cites a production gap: 51% of open-model deployments reach production, compared with 63% for closed models.

    Combined views

    43.7K

    17 Sources, first seen 21d ago

    Combined views

    43.7K

    17 Sources, first seen 21d ago

    176 likes
    21d ago
    first seen 21d ago
    176 likes34 comments60 saves64 reposts
    34 comments
    60 saves
    64 reposts

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.

    Today's Rank

    —

    Not ranked yet

    Today's Rank

    —

    Not ranked yet

    17 Sources

    Rohan Paul@rohanpaul_aiMozilla just published a 91-page report and it says open-weight AI is now only about 4 months behind the frontier. - 8 of OpenRouter's 10 most-used models by August token volume were open-weight, and 7 were Chinese-built, while DeepSeek became the first open model to lead the platform in weekly requests. - However, the economics are almost upside down. Open models handled roughly 20% of measured OpenRouter usage but captured only about 4% of model-layer revenue in the cited 2025 window. Mozilla attributes much of that mismatch to pricing, with closed models costing roughly 6x more per call at about 90% capability parity. The comparison shows why high usage does not necessarily translate into high revenue when one class of models is dramatically cheaper. Mozilla also warns that the revenue measurement is older than its 2026 usage data, so the current revenue split may be different. - Open models are spreading faster than they reach production: 79% of surveyed developers use them, but only 51% of open-model deployments reach production versus 63% for closed models. - DeepSeek showed that a new pretraining run may no longer be necessary for a major capability jump, gaining roughly 10 index points and then another 8 through post-training passes. -Even model diversification may provide less protection than assumed: Kimi K3 and Claude Fable 5 had a 0.72 per-task failure correlation, meaning supposed backup models often fail on the same problems.21d
    Jonathan Richard Schwarz@schwarzjn_Two charts that should make you optimistic about AI sovereignty: Open-weight models are closing the gap with the frontier, and reaching it is getting easier, with a record number of near-frontier models released in 2026 from a wide range of institutions. In addition, you don't even need to compete for max performance: ~75% of model calls to OAI/Anthropic go to models chosen for a favourable cost/performance trade-off, not raw capability. h/t @FT @OpenRouter16d

    17 Sources

    Rohan Paul@rohanpaul_aiMozilla just published a 91-page report and it says open-weight AI is now only about 4 months behind the frontier. - 8 of OpenRouter's 10 most-used models by August token volume were open-weight, and 7 were Chinese-built, while DeepSeek became the first open model to lead the platform in weekly requests. - However, the economics are almost upside down. Open models handled roughly 20% of measured OpenRouter usage but captured only about 4% of model-layer revenue in the cited 2025 window. Mozilla attributes much of that mismatch to pricing, with closed models costing roughly 6x more per call at about 90% capability parity. The comparison shows why high usage does not necessarily translate into high revenue when one class of models is dramatically cheaper. Mozilla also warns that the revenue measurement is older than its 2026 usage data, so the current revenue split may be different. - Open models are spreading faster than they reach production: 79% of surveyed developers use them, but only 51% of open-model deployments reach production versus 63% for closed models. - DeepSeek showed that a new pretraining run may no longer be necessary for a major capability jump, gaining roughly 10 index points and then another 8 through post-training passes. -Even model diversification may provide less protection than assumed: Kimi K3 and Claude Fable 5 had a 0.72 per-task failure correlation, meaning supposed backup models often fail on the same problems.21d
    Jonathan Richard Schwarz@schwarzjn_Two charts that should make you optimistic about AI sovereignty: Open-weight models are closing the gap with the frontier, and reaching it is getting easier, with a record number of near-frontier models released in 2026 from a wide range of institutions. In addition, you don't even need to compete for max performance: ~75% of model calls to OAI/Anthropic go to models chosen for a favourable cost/performance trade-off, not raw capability. h/t @FT @OpenRouter16d

    Sentiment

    Positive——Negative

    Summary

    Not enough discussion yet.

    No sentiment analysis available yet.