Blog
Latest articles
4 min readvLLM: The Fastest Open Source LLM Inference Server
PagedAttention gives vLLM 2-24x throughput over naive implementations. Here is how to set it up, configure batching, quantize models, and calculate hardware costs.
10 min readFine-Tuning an LLM with QLoRA on a Single GPU
QLoRA makes fine-tuning 70B models accessible on a single consumer GPU. Here is the complete setup guide for fine-tuning Llama 3 with Unsloth.
8 min readMLflow for Experiment Tracking: Setup, Usage, and When It Is Enough
MLflow tracks parameters, metrics, and artifacts for ML experiments. Here is how to set it up in 5 lines, integrate with training loops, and when to use it over W&B.
9 min readDSPy: Automatic Prompt Optimization for Complex LLM Pipelines
DSPy optimizes LLM prompts automatically using your data. Here is when it helps, when it does not, and a complete setup guide for a real use case.
9 min readCrewAI: Building Multi-Agent Systems in Python
CrewAI lets you define agents with roles, assign them tasks, and have them collaborate. Here is when multi-agent beats single-agent, and when it does not.
5 min readOpen Source LLM Benchmarks 2026: How They Compare to GPT-4o
Llama 3.3 70B scores ~87% on MMLU versus GPT-4o at ~88.7%. The gap is closing. Here is where open source wins, where it still loses, and what the benchmarks actually measure.
16 min readCutting LLM API Costs by 50%+: Every Technique That Works in 2026
Six proven techniques to reduce your LLM API spend. Real pricing numbers, a startup case study reducing from $800 to $320/month, and specific implementation guidance.
8 min readLLM API Pricing Comparison 2026: Every Major Model, Real Numbers
Complete LLM API pricing table with per-request cost calculations. Which model is cheapest for coding, summarization, and classification? Real numbers, no estimates.
7 min readPrompt Caching With Anthropic and OpenAI: How to Cut Costs by Up to 90%
How prompt caching works on Anthropic and OpenAI, when it saves money, and how to implement it. Real cost reduction numbers with code examples.
9 min readModel Routing: How to Cut LLM Costs 50-70% Without Sacrificing Quality
Model routing automatically sends simple queries to cheap models and complex ones to expensive models. With GPT-4o-mini at $0.15/1M tokens vs GPT-4o at $2.50/1M, the savings are substantial.
8 min readSemantic Caching: How to Serve LLM Responses Without Calling the API
Semantic caching stores LLM responses and returns them when a new query is semantically similar to a cached one. In customer support applications, hit rates of 15-40% are realistic.
5 min readOpenAI Batch API: Get 50% Off for Non-Real-Time Requests
OpenAI's Batch API cuts costs by 50% for any request that can wait up to 24 hours. If you have data labeling, nightly analysis, or content moderation workloads, you should be using it.