Microsoft Phi-4
Microsoft's 14B dense transformer that consistently beats models 5x its size on MATH and GPQA, available under the MIT license for unrestricted commercial use.
James Kowalski
AI Benchmarks & Tools Analyst
James is a software engineer turned tech writer who spent six years building backend systems at a fintech startup in Chicago before pivoting to full-time analysis of AI tools and infrastructure. His engineering background means he doesn't just read the spec sheet - he runs the benchmarks, profiles the latency, and checks whether the marketing claims hold up under real workloads.
He studied Computer Science at the University of Illinois at Urbana-Champaign, where he first got hooked on natural language processing during a senior research project on sentiment analysis. He later completed a certificate in data journalism from Northwestern's Medill School.
At Awesome Agents, James owns the leaderboards and tool comparison coverage. He maintains the site's benchmark tracking methodology and is the person who actually runs the numbers before publishing any ranking. He is also an open-source advocate and contributes to several projects in the LLM inference space.
Based in Chicago, IL.
Articles by James Kowalski
Mistral Large 3
Mistral Large 3 is a 675B-parameter MoE model activating 41B per token with native multimodal support, a 256K context window, and Apache 2.0 licensing - Europe's first frontier-class open-weight model.
Mistral Small 3.2
Mistral Small 3.2 is a 24B dense model with strong function calling, multimodal vision, and 128K context under Apache 2.0 - optimized for production tool-use pipelines and EU-compliant deployments.
Nemotron 3 Nano 30B-A3B
NVIDIA's hybrid Mamba2+MoE model packs 31.6B total parameters but activates only 3.2B per token, delivering frontier-class reasoning with 3.3x the throughput of comparable models on a single H200 GPU.
Qwen3.5-122B-A10B vs DeepSeek V3.2: Efficiency vs Raw Power in Open-Weight AI
A benchmark-by-benchmark comparison of Qwen3.5-122B-A10B and DeepSeek V3.2 - the efficiency-optimized underdog versus the brute-force open-source heavyweight.
Qwen3.5-122B-A10B vs Llama 4 Maverick: The Efficiency Gap Nobody Expected
A data-driven comparison of Alibaba's Qwen3.5-122B-A10B and Meta's Llama 4 Maverick - two open-weight MoE models with radically different approaches to parameter efficiency and benchmark performance.
Qwen3.5-122B-A10B vs Mistral Large 3: When 4x More Parameters Buys You Less
A data-driven comparison of Qwen3.5-122B-A10B and Mistral Large 3 - two Apache 2.0 MoE models where the smaller one dominates text benchmarks despite a 4x active parameter disadvantage.
Qwen3.5-27B vs Gemma 3 27B: Same Parameter Count, Completely Different Models
A data-driven comparison of Alibaba's Qwen3.5-27B and Google's Gemma 3 27B - two 27B dense models that share a parameter count and almost nothing else.
Qwen3.5-27B vs Mistral Small 3.2: Apache 2.0 Heavyweights Go Head to Head
A data-driven comparison of Alibaba's Qwen3.5-27B and Mistral's Small 3.2 - two Apache 2.0 dense models in the 24-27B range with very different benchmark profiles and deployment strengths.
Qwen3.5-27B vs Phi-4: When Twice the Parameters Is Not Twice as Obvious
A data-driven comparison of Alibaba's Qwen3.5-27B and Microsoft's Phi-4 - a 27B hybrid architecture versus a 14B STEM specialist, testing whether raw parameter count or training efficiency wins in practice.
Qwen3.5-35B-A3B vs GLM-4.7-Flash: Two Chinese MoE Models, Very Different Strengths
Head-to-head comparison of Qwen3.5-35B-A3B and GLM-4.7-Flash - two Chinese-origin 30B-A3B MoE models with Apache 2.0/MIT licenses that dominate different benchmarks despite near-identical parameter budgets.
Qwen3.5-35B-A3B vs Llama 4 Scout: 3B Active Parameters vs 17B - Does 5.7x More Compute Actually Win?
David vs Goliath: Qwen3.5-35B-A3B activates 3B parameters and beats Llama 4 Scout's 17B active on MMLU-Pro, GPQA, and coding benchmarks - but Scout's 10M context window and native multimodal support tell a different story.
