Hands-on comparison of Gemini 3.1 Pro, Claude Opus 4.6 and GPT-5.3 Codex

On October 16, 2026, DEV Community's AI section published a comparison of three coding models released in February 2026: Gemini 3.1 Pro, Claude Opus 4.6, and GPT-5.3 Codex. The article's central conclusion is that no single model dominates every use case, with each one standing out in a different area.

The numbers cited are as follows. Gemini 3.1 Pro scored 77.1% on ARC-AGI-2 and leads on algorithmic coding, at a lower price of $2 per million input tokens and $12 per million output tokens, which the article describes as 7x lower cost. Claude Opus 4.6 reached 80.8% on SWE-Bench Verified, the highest among the three on real-world coding tasks, and includes a distinctive Agent Teams feature. GPT-5.3 Codex dominates terminal workflows, posting 77.3% on Terminal-Bench 2.0.

The article frames February 2026 as the month AI labs shifted from competing on benchmarks to competing on developer experience, though it does not detail how each model was evaluated or what tasks were used. As of this report, the story remains at the level of a single comparative review; the material contains no release dates for the three models, no vendor responses, and no follow-up updates.