Blog
RSSDeepSeek V4.1 Flash 1017: the most-called LLM, reviewed with data
OpenRouter's weekly data shows DeepSeek V4.1 Flash 1017 as the most-used LLM at 56.5T tokens. What the numbers do and do not say, plus live prices.
Prompt caching explained: how cache-hit pricing cuts your LLM bill
Cache read tokens cost a fraction of normal input tokens, and the fraction differs a lot by model. Live prices, a worked example and how to get more hits.
Who pays for AI: what a16z's spending data means for API buyers
a16z's October 2026 data shows consumer AI spend concentrated in a few heavy users. What that means when choosing between plans and pay-per-token.
LLM API vs chat subscription: when pay-per-token makes sense
A flat monthly plan and a per-token API solve different problems. How to tell which is cheaper for you, with a break-even check and sourced data.
DeepSeek, Qwen and GLM API pricing compared (per million tokens)
Per-million-token prices for DeepSeek V4.1 Flash 1017, Qwen3.7 and Qwen3.8 Max, GLM-5.3 and MiniMax-M3, read live from the ModelSite catalog.
Call DeepSeek, Qwen and GLM with one OpenAI-compatible API key
Use one API key and the OpenAI SDK to call DeepSeek, Qwen and GLM. Change the model name, keep your code. Includes curl, Python and live prices.