OpenAI Launches GPT-5.2 Amid Google’s “Code Red” Alert

Bassyonni

ago 19 minutes

OpenAI Launches GPT-5.2 Amid Google’s “Code Red” Alert

OpenAI has unveiled GPT-5.2, marking its third significant model release since August. This debut comes as tech giant Google raises alarms with its “Code Red” alert due to increased competition in the artificial intelligence sector.

Improvements of GPT-5.2

GPT-5.2 rolls out enhancements targeting user experience and performance. Following the July launch of GPT-5, which introduced a balance between quick responses and simulated reasoning, GPT-5.1 added eight preset personality options for a more engaging interaction. Despite these upgrades, users found earlier versions somewhat unapproachable.

Performance Benchmarks

Although GPT-5.2 aims to compete with models like Gemini 3, OpenAI has not provided direct performance comparisons on its promotional site. Instead, it highlights improvements over earlier models and its results on the newly introduced GDPval benchmark.

GPT-5.2 Thinking: Scored 55.6% on the SWE-Bench Pro, surpassing Gemini 3 Pro (43.3%) and Claude Opus 4.5 (52.0%).
On GPQA Diamond: Achieved a score of 92.4%, slightly ahead of Gemini 3 Pro’s 91.9%.

Task Completion Efficiency

GPT-5.2 reportedly outperforms human professionals on 70.9% of tasks in the GDPval benchmark, contrasting with Gemini 3 Pro’s 53.3%. Notably, it executes these tasks more than eleven times faster, requiring less than 1% of the cost associated with human experts.

Improved Reliability

The new model also claims a 38% reduction in confabulations, attributed to advancements in its training. OpenAI’s post-training lead remarked that GPT-5.2 “hallucinates substantially less” than previous versions. Despite these claims, benchmarks should be approached cautiously, as independent evaluations will take time to emerge.

Final Thoughts

For users relying on ChatGPT for professional tasks, expect noticeable enhancements in capabilities. The integration of better coding performance and incremental upgrades signifies OpenAI’s continuous commitment to optimizing its AI offerings amid growing competition.