Coinbase fine-tuned Qwen3.5-9B on their internal fraud data and got better results than Claude Opus 4.5 on fraud detection, with 55% faster latency and at a fraction of the cost. The takeaway: smaller models trained on your own data outperform frontier models on narrow, specialized tasks.

• Qwen3.5-9B fine-tuned on Coinbase's fraud data outperformed Claude Opus 4.5 on their private benchmark, with latency dropping from 1.515 to 0.683 seconds
• Live A/B testing showed a fraud detection agent using the fine-tuned model reduced fraudulent transactions by 30% and fraud value by 22%, saving about 3x the inference cost
• Frontier models scored lower on recall and F1 for this specific task, suggesting specialized models beat generalists on narrow high-stakes problems

Together AI walks through building and deploying a Jev-like classification model on Qwen 3.5 4B for $17. The guide covers dataset selection, fine-tuning, and deploying the model behind an HTTP endpoint so you can query it for tasks like intent detection or sentiment analysis.

• You can fine-tune a working classification model on 38,000 examples for roughly $17 and have it deployed in 25 minutes using Together's platform
• The model accepts JSON-formatted state plus predefined questions and returns structured answers (scores, booleans, or multiple choice), making it practical for e-commerce returns, paper categorization, or sentiment rating
• The entire process is automated through provided scripts—you clone a repo, download datasets, run a training command, and get back a deployed API endpoint ready to integrate

This guide shows how to use Apple Silicon and the Gemma 3 270M model to fine-tune a small language model offline in under 10 minutes. It walks through installing the uv/MLX toolchain, preparing a simple code-review dataset, and running the training on any M-chip Mac without a cloud GPU.

• Gemma 3 270M runs on ~830MB RAM and hits 150+ tokens/sec on an M3, fine-tuning in under 10 minutes via MLX's uv toolchain (as fast as 3 minutes on an M1 Air)
• A dataset of only ~60 Python code snippets paired with expert review notes was enough to teach the model to flag security issues (SQL injection, unclosed files, insecure shell commands) and style problems
• The entire workflow runs offline on Apple Silicon with no cloud GPU or rental fees required

ConvApparel is a human-AI conversation dataset and evaluation framework designed to quantify and reduce the realism gap in LLM-based user simulators. Using dual-agent shopping dialogues and three validation pillars—statistical alignment, human-likeness scoring, and counterfactual tests—researchers show that data-driven simulators adapt more like real users but still exhibit detectable artifacts.

• Google researchers built ConvApparel by secretly routing users to a "Good" or "Bad" recommender across 4,000+ sessions and 15,000 turns, capturing real turn-by-turn satisfaction/frustration ratings.
• They tested prompting, in-context learning, and fine-tuning (on Gemini 2.5 Flash) as user simulators, validating each against stats alignment, a human-likeness discriminator, and counterfactual "Good vs Bad" agent tests.
• Prompt-based and in-context simulators failed to show genuine frustration when switched to the "Bad" agent, staying too smooth and helpful.
• Only the fine-tuned simulator closely matched real humans' shifts in frustration and satisfaction, showing targeted training on real interactions beats prompting or few-shot examples for realistic simulation.

Organizations are increasingly faced with the decision of whether to implement Retrieval-Augmented Generation (RAG) or fine-tuning for their AI initiatives. RAG connects large language models to external databases, allowing access to real-time information, reducing inaccuracies, and enhancing security and traceability. However, implementing RAG comes with its own technical challenges that require careful planning and maintenance.

• RAG connects LLMs to external data sources so responses reflect real-time information rather than static training data
• Grounding responses in retrieved data cuts down on hallucinations and improves accuracy
• RAG offers added benefits like source traceability and better security for proprietary data
• Setting up RAG requires substantial infrastructure work—data pipelines, indexing, and ongoing maintenance for quality and scalability