/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Smaug-72B, a Qwen-72B-based open-source LLM released by Abacus AI, tops the Hugging Face Open LLM leaderboard and outperforms GPT-3.5 on several benchmarks

A new open-source language model has claimed the throne of the best in the world, according to the latest rankings from Hugging Face …

VentureBeat Michael Nuñez

Context & Ripple Effects

Smaug-72B is an Abacus AI release built on Qwen-72B, making its leaderboard result an early example of a derivative open model gaining visibility through benchmark performance rather than a wholly new base architecture.

Later coverage shows the underlying Qwen family remained central to open-model competition: Qwen led the updated Open LLM Leaderboard, while Alibaba subsequently released an open reasoning model positioned on lower compute needs. That arc makes Smaug-72B's result relevant as evidence that open-weight ecosystems can compound improvements around a base model.

First-order effects

  • Abacus AI gains a credible public performance signal for Smaug-72B after it reaches the top of the Hugging Face Open LLM leaderboard and exceeds GPT-3.5 on several reported benchmarks.
  • Developers evaluating open models have another high-ranking Qwen-72B-derived option, while GPT-3.5 becomes a more explicit comparison point for this class of model.

Second-order effects

  • The result raises the incentive for model builders to fine-tune and specialize strong open base models, competing on evaluation results and task fit instead of only training new foundational models.
  • Leaderboard standing becomes more consequential for open-model adoption, but comparisons will increasingly depend on the scope and design of the benchmarks used—a dynamic reflected in the later six-benchmark leaderboard update.

Third-order effects

  • If derivative models repeatedly approach or surpass older proprietary benchmarks, value can shift toward the open-weight complement economy: tuning, deployment, evaluation, and support around accessible base models.
  • The durable competitive question becomes less whether a model is open or closed in isolation and more whether its surrounding ecosystem can translate benchmark gains into reliable use cases; benchmark leadership alone does not establish that outcome.

The trend: Open-weight model ecosystems are turning capable base models into platforms for rapid derivative innovation and increasingly visible benchmark competition.

Discussion

  • @bindureddy Bindu Reddy on x
    Here is the link to the HF leaderboard with Smaug on top! https://huggingface.co/... [image]
  • @gblazex Blaze on x
    Amazing! New Qwen 1.5 72B model rivals GPT-4 even on the Length Adjusted Alpaca v2 Leaderboard. (original alpaca has heavy length bias, this gives a clearer picture) Thank you @jeremyphoward for pointing me to the lengths. Also idea for length adjustment from @teortaxesTex [image…
  • @bindureddy Bindu Reddy on x
    Smaug-72B - The Best Open Source Model In The World - Top of Hugging LLM LeaderBoard!! Smaug72B from Abacus AI is available now on Hugging Face, is on top of the LLM leaderboard, and is the first model with an average score of 80!! In other words, it is the world's best... [image…
  • @scobleizer Robert Scoble on x
    First model with average score of 80 on @huggingface. Open source coming on strong from @abacusai