Archive · Theme
Infra and cost
Inference, GPUs, price per token, and what changes when cost drops tenfold.
GLM-5.3-Flash: an MIT-licensed 320B model served entirely on Chinese chips, at US$ 0.15 per million tokens
It is the clearest datapoint yet that a competitive model can be trained and served at scale without NVIDIA hardware — and it prices Opus-class agentic work at cents.
The government published the R$ 959 million tender for the Rio Grande do Norte AI supercomputer — bids due 8 October
It is the first concrete, dated and funded step toward sovereign AI compute in Brazil — and the tender terms will shape which stacks Brazilian engineers get to use.
A large context window does not mean the context is used: models lose what sits in the middle
It explains why “dump everything in the context” fails and why RAG and evidence ordering still matter.
DeepSeek-V3: a 671B model trained in 2.79 million H800 hours, without a single loss spike
It was the report that made the market rethink what a frontier model costs to train — and the origin of several techniques 2026 open models still use.
PagedAttention: treating the KV cache like virtual memory doubled inference throughput — and gave birth to vLLM
Cost per token is largely memory cost; this is the paper that explains why.