Malaysian workers are already further ahead on artificial intelligence than the companies that employ them, and OpenAI's newest release lands right on that gap. On Thursday the company launched GPT-5.6, a family of models it is selling on being cheaper and more efficient to run, alongside a new tool built for office teams.
Editor
Kai T chevron_right
Table of Contents
What OpenAI actually launched
GPT-5.6 comes in three versions, according to TechCrunch: Sol, the top-end workhorse, Terra in the middle, and Luna as the budget option. OpenAI says the models are more token efficient than before, and chief executive Sam Altman told CNBC that Sol is 54 percent more token efficient on coding tasks. Since AI services bill by the token, efficiency is really a price story.
The company also released ChatGPT Work, a companion for enterprise teams that runs on desktop, web, and mobile and helps with everyday clerical jobs like drafting documents, spreadsheets, and presentations. OpenAI calls 5.6 its strongest cybersecurity model yet, tuned for defensive work such as threat modelling, code review, patching, and testing your own systems for weaknesses before attackers find them. GPT-5.6 is available now across ChatGPT, the Codex coding tool, and the OpenAI API.
Pricing tells the real story. Per million tokens, OpenAI lists Sol at 5 dollars input and 30 dollars output, Terra at 2.50 and 15 dollars, and Luna at 1 dollar and 6 dollars. Much of the marketing aims at rival Anthropic: OpenAI cites the Artificial Analysis Coding Agent Index to claim Sol beats Anthropic's Fable 5 while using less than half the tokens and costing about a third less.
Why this matters for Malaysia
Malaysia has a particular AI shape that these changes speak to. Microsoft's 2026 Work Trend Index found 78 percent of Malaysian knowledge workers use AI at work at least weekly, well above the 58 percent global average, and 24 percent count as the most advanced "frontier" users against 16 percent globally. An EY 2025 survey put GenAI use among Malaysian employees at 93 percent. The workers here are not the bottleneck.
The organisations are. The same Microsoft research found 67 percent of Malaysian IT budgets go to keeping existing systems running, leaving under 15 percent for anything new, while 61 percent of AI users had not told their manager they were using the tools at all. That is a market where cheaper per-token pricing, a ready-made app in ChatGPT Work, and a model pitched on security could turn quiet, unsanctioned AI use into something a finance or IT lead can actually approve.
The bigger picture
Cost has been the quiet brake on Malaysia's AI ambitions. The country is spending heavily to become an "AI nation," including a two billion ringgit sovereign AI cloud under Budget 2026, while local developers already lean on tools like Anthropic's Claude Code and Google's Gemini. A price war between OpenAI, Anthropic, and Meta, all of which shipped new models this week, drags the cost of frontier capability down for everyone paying in ringgit. Good news for a startup counting tokens, and a fresh headache for anyone trying to pick one stack and commit.
GPT-5.6 will not close the gap between eager Malaysian workers and cautious Malaysian employers on its own. But cheaper models paired with a security story is close to what a nervous boss has been waiting for.
Images courtesy of Chris Ried and Vitaly Gariev on Unsplash.