Stories from the gateway
Fundamentals
Inference vs Training Cost: Which One Is Actually Your Problem
Training is a one-off capital cost you never pay. Inference is a recurring cost you pay per request. How the two differ and why the distinction shapes pricing.
· 9 min read
Fundamentals
Instruction Tuning: How a Text Predictor Becomes an Assistant
A pretrained model continues text; it does not answer questions. Instruction tuning is the small, cheap stage that turns one into the other.
· 8 min read
Fundamentals
KV Cache Explained: The Memory Behind Long Context
· 8 min read
Fundamentals
Layer Normalisation: The Part That Keeps Training Stable
· 8 min read
Fundamentals
Load Balancing in MoE: Stopping Experts Going Idle
· 9 min read
Fundamentals
Logits and Softmax: What a Model Really Outputs
· 8 min readFundamentals
Long Context Degradation: How Quality Falls Before the Limit
· 9 min read
Fundamentals
max_tokens and Truncation: Why Your Output Stops Mid-Sentence
· 8 min read
Fundamentals
Model Parallelism: How Huge Models Are Split to Run
· 9 min readFundamentals
Model Size vs Capability: Why Bigger Stopped Meaning Better
· 9 min readFundamentals
Multi-Head Latent Attention Explained
· 9 min readFundamentals
Needle-in-a-Haystack Tests: What They Prove and What They Miss
· 9 min readFundamentals
Paged Attention: Virtual Memory for the KV Cache
· 9 min readFundamentals
Planning Storage for Self-Hosted Model Weights
· 9 min readReady to start?
Every model, one key. Start free and upgrade when you are ready.