Computerworld finds open models rising

Computerworld reported on May 19 that enterprise technology buyers are increasingly looking beyond the biggest proprietary AI systems and testing smaller, open models for narrower business jobs. The publication said IT decision-makers are weighing lower cost, easier customization and tighter deployment control against the broader capabilities offered by closed, frontier systems. The shift is playing out as large software vendors keep embedding more powerful default models into workplace products. GitHub, for example, said on May 17 that GPT-5.3-Codex became the base model for Copilot Business and Copilot Enterprise, replacing GPT-4.1. ### Why are companies looking at smaller open models now? Computerworld said smaller, open models are gaining traction because enterprise buyers increasingly want systems that fit a defined task rather than a broad, general-purpose chatbot role. The report said those models are being evaluated as cheaper to run and easier to adapt to internal workflows than large proprietary offerings. (computerworld.com) Agam Shah wrote that large language models remain popular, but IT decision-makers are trying out open alternatives that may better match enterprise needs. That framing puts cost, customization and task fit at the center of procurement decisions rather than model size alone. ### What are buyers weighing against those benefits? (computerworld.com) GitHub said GPT-5.3-Codex became the default base model for Copilot Business and Copilot Enterprise on May 17. The company said the change applies to those paid organizational plans and replaces GPT-4.1 as the default option used when an organization has not selected another supported model. The contrast is that hyperscalers and large platform vendors are still moving larger or newer frontier systems directly into software that companies already use. That means enterprise customers may receive more capable default models inside existing contracts even as some technology teams look for smaller, more controllable tools for specific workloads. That is an inference from the Computerworld report and GitHub’s product change. (github.blog) ### Is this a rejection of large proprietary models? Computerworld did not describe the move as an abandonment of proprietary systems. The report said large models such as ChatGPT and Gemini remain popular even as the market opens up to smaller and more open options. That leaves enterprises comparing different strengths. (computerworld.com) Open models can offer more room for tuning, deployment flexibility and cost control, while closed models bundled into major software products may offer stronger out-of-the-box performance or easier adoption through existing vendor relationships. That comparison is drawn from the Computerworld report and GitHub’s changelog. ### What does GitHub’s move show about the other side of the market? GitHub said on March 18 it had announced long-term support and base-model changes for Copilot, and on May 17 it switched Copilot Business and Enterprise organizations to GPT-5.3-Codex. The company also said GPT-4.1 will remain force-enabled at a 0x multiplier for the time being but will deprecate alongside the launch of usage-based billing on June 1, 2026. (computerworld.com) Those dates show that large vendors are still standardizing customers onto newer flagship models inside enterprise software. At the same time, Computerworld’s reporting indicates some buyers are making a separate calculation about whether every use case needs that class of model. That is an inference based on the two published updates. (github.blog) ### What happens next? June 1, 2026 is the next concrete date in this market shift. GitHub said that is when usage-based billing launches and GPT-4.1 deprecates for Copilot Business and Enterprise, giving enterprise customers another point at which model choice, pricing and default deployment settings may be reassessed. (github.blog) (computerworld.com)