ModelsGLM-5V Turbo: Z.ai's Vision Model for Screenshots and GUI Agents
GLM-5V Turbo reads images, video and files and answers in text, tuned for design-to-code and GUI agents. What is published, what is not, and how to evaluate it.
ReadNotes on models, agents, and what inference costs
Practical writing for developers building with large language models — how they work, how to pick one, and how to keep the bill predictable.
Picking between models — and testing them on your own work.
Models
ModelsGPT Audio Guide: Speech In, Speech Out Over Chat Completions
gpt-audio takes and returns audio through Chat Completions. How audio tokens are priced, what the endpoint does not support, and when a separate pipeline wins.
Read
ModelsGPT Audio Mini: Speech In, Speech Out, at a Third of the Price
GPT Audio Mini takes and returns audio over Chat Completions for far less than GPT Audio. What it costs, what it cannot do, and when a chained pipeline is better.
Read
ModelsGPT-3.5 Turbo 16k: A Model Name That Outlived Its Reason
The 16k variant once bought four times the context of GPT-3.5 Turbo. The base model now has the same window at a sixth of the input price. What that means for you.
Read
ModelsGPT-3.5 Turbo Guide: The 23 October Shutdown and How to Move
OpenAI retires gpt-3.5-turbo on 23 October 2026. What the model was, which snapshots and aliases go with it, and a migration plan that fits in a week.
Read
ModelsGPT-3.5 Turbo Instruct: Porting the Last Completions Model
OpenAI switched off gpt-3.5-turbo-instruct on 28 September 2026. What made a completions model different, and how to port its prompts to a chat model.
ReadModelsGPT-4 in 2026: What It Was, What It Costs and Moving Off It
The original GPT-4 still answers on an 8K window at $30 and $60 per million tokens, until 23 October 2026. Why code still calls it and how to plan the migration.
ReadModelsGPT-4 Turbo Guide: 128K In, 4K Out, and a 23 October Deadline
GPT-4 Turbo brought a 128K window and vision to GPT-4 at a third of the input price. Why its 4,096-token output cap matters and what replaces it after 23 October.
ReadModelsGPT-4.1 Guide: What It Did Well, and Its January 2027 Exit
GPT-4.1 brought a 1M-token window and stronger instruction following to OpenAI's API in 2025. Its shutdown is now set for 14 January 2027. What to know and do.
ReadModelsGPT-4.1 Mini Guide: The Cheap Workhorse and Its Cost Maths
GPT-4.1 mini offers the full model's 1M window at a fifth of the price. Which jobs it suits, the batch and cache arithmetic, and its 14 January 2027 retirement.
ReadModelsGPT-4.1 Nano: The 1M-Token Small Model and Its October Shutdown
GPT-4.1 nano paired a million-token window with the lowest price in its family. What it was good at, how it compares with GPT-4o mini, and how to migrate off it.
ReadModelsGPT-4o Guide: Three Snapshots, Three Prices, Two Shutdown Dates
GPT-4o is one name over three dated snapshots with different prices, output limits and retirement dates. Which one your code calls, and what to do before each deadline.
ReadShowing 85–96 of 446 articles