@nathanhabib1011: best models < 128B params on SWE-bench_pro... @Alibaba_Qwen 3.6 27b still crazy, closely followed by @ornith_ 35B
Summary
Tweet highlighting top-performing AI models under 128B parameters on the SWE-bench_pro benchmark, noting Alibaba Qwen 3.6 27B and ornith 35B as leading contenders.
View Cached Full Text
Cached at: 07/01/26, 08:05 AM
best models < 128B params on SWE-bench_pro… @Alibaba_Qwen 3.6 27b still crazy, closely followed by @ornith_ 35B https://t.co/9BmWE8WGw1
Similar Articles
@Sentdex: For anyone who isn't sure, this is how you release a model and talk about the performance. Not 3-5 cherry-picked benchm…
A tweet by Sentdex highlights Alibaba Qwen's transparent benchmark reporting for the Qwen3.7-Max model, contrasting it with others who cherry-pick benchmarks.
@LottoLabs: There’s so much demand for a good small model, look at top downloaded qwen models All < 9b
Observation that there is high demand for small AI models, as seen in the top downloads of Qwen models under 9B parameters.
Qwen3.6-27B
Alibaba's Qwen team released Qwen3.6-27B, a new 27-billion-parameter language model, accompanied by benchmark results.
A 4b model is now beating 30b ones at web research and the reason is not size
A 4 billion parameter open model from the Apodex family outperforms 30 billion parameter models on web research benchmarks, attributed to careful training data and self-verification techniques rather than raw scale, suggesting a more democratic trajectory for AI capability.
Is this true? Alibaba has released Qwen3.8, a new 2.4T parameter model that they claim is behind only Claude Fable 5 in performance.
Alibaba released Qwen3.8, a 2.4 trillion parameter model claiming performance second only to Claude Fable 5.