New AI Models Drop: GPT-5.4 & Gemini 3.1 Variants
The AI arms race is accelerating with fresh model drops including GPT-5.4, which boasts sharper reasoning capabilities. Google has also released several new Gemini 3.1 variants, continuing the rapid pace of innovation and competition between major AI labs. These releases signal a new wave of performance improvements for developers and researchers.
OpenAI's GPT-5.4 boasts a 1.05 million token context window and unifies the company's previous Codex and GPT lines into a single model. The API is priced at $2.50 per million input tokens and $15.00 per million output tokens, with a more powerful Pro version also available. Google's Gemini 3.1 release includes a suite of models tailored for specific tasks, such as the cost-efficient Flash-Lite for high-volume use cases and Deep Think for complex research problems. The Pro version is a frontier reasoning model with a 1 million token context window and advanced multimodal capabilities, able to process text, images, video, and up to 8.4 hours of audio in a single prompt. On performance benchmarks, GPT-5.4 scored 92% on GPQA for graduate-level reasoning, while Gemini 3.1 Pro Preview leads in visual reasoning tests (82%) and shows significantly lower hallucination rates. The competition remains fierce, with models like Anthropic's Claude Opus 4.6 still considered a top performer for complex, multi-file coding tasks. This technological escalation is directly impacting software engineering careers, creating a
