Tracked topic
Qwen3 is an AI model family developed by Alibaba, released as a set of large language models for natural-language tasks.
…Qwen kept inventing the workload it was optimizing for, and Gemma kept pointing out that there was no workload, no benchmark, and no reason for any of this. "You are optimizing for…
…Qwen3-Coder has consistently topped my benchmarks I couldn't find a better local model for Python development If you've spent any amount of time experimenting with local coding models over…
…Model Ollama Tag Size on Disk GPT-OSS gpt-oss:20b 13GB Devstral Small 2 devstral-small-2:latest 15GB Qwen3-Coder qwen3-coder:30b 18GB Qwen3-Coder 30B was the fastest…
…Qwen 3.5's thinking mode is inherited and refined from Qwen 3, so it's more battle-tested. And the 9B actually beats the previous-gen Qwen3-30B on reasoning benchmarks…
…Meta's own benchmarks claim it beats Gemma 2 2.6B and Phi-3.5 mini, though obviously Meta would say that. I picked it because the Llama family is the default…
…When purpose-built coding models started appearing online with benchmark scores that piqued my interest, I decided to try some out. The model I've stuck with is Qwen2.5-Coder 32B…
…To find out which one deserved the spot the most, I tested Meta's Llama Scout 17B, Qwen3-Coder 30B and Gemma 4 26B against one another by asking them to build…
…I'd seen the Qwen family come up a lot, but every time I looked into it, the conversation was always about coding benchmarks and developer workflows. So I kind of just…
…Qwen 3.6 keeps its attention memory from growing at all Gated DeltaNet is incredibly interesting Qwen's recent models, including the Qwen 3.6 family of models, use Gated DeltaNet, a…
…You can find benchmarks and capability breakdowns all over the place - LM Studio even tells you what each model is optimized for in the description. But I wanted to see it for…