Intel / article

Openai Gpt5 5 Is The New Leading AI Model

Source: artificialanalysis.ai
Author: Artificial Analysis articles
Date: Published Apr 23, 2026

Context

OpenAI's GPT-5.5 tops the Artificial Analysis Intelligence Index by 3 points, ending the three-way tie between Anthropic, Google and OpenAI (artificialanalysis.ai). At its highest reasoning effort, xhigh, it leads Terminal-Bench Hard, GDPval-AA and APEX-Agents-AA, and trails only other OpenAI models on CritPt and AA-LCR; it comes second to Gemini 3.1 Pro Preview on three more. On GDPval-AA, Artificial Analysis's test of economically valuable real-world tasks built on OpenAI's GDPval dataset, it reaches an Elo of 1785, about 30 points ahead of Claude Opus 4.7 and about 470 ahead of Gemini 3.1 Pro Preview.

Per- pricing doubled from GPT-5.4 to $5 per million input tokens and $30 per million output tokens. GPT-5.5 used about 40% fewer output tokens to run the index, which absorbed most of the increase, so the net cost of a full run rose about 20%. Artificial Analysis says it is still about 30% cheaper to run than Opus 4.7 at max effort.

OpenAI gave Artificial Analysis pre-release access to all five reasoning effort settings, from xhigh down to non-reasoning, which control how much the model thinks before answering. Artificial Analysis calls them a clear ladder for trading intelligence against cost. GPT-5.5 at medium matches Opus 4.7 at max on the index for about a quarter of the cost, roughly $1,200 against $4,800, although Gemini 3.1 Pro Preview reaches the same score for about $900. At low, it approximates non-reasoning Opus 4.7 at about half the cost.

The largest gains over GPT-5.4 came on AA-Omniscience, up 14 points, and τ²-Bench Telecom, a customer service test, up 7. AA-Omniscience rewards factual knowledge and punishes . GPT-5.5 has the highest accuracy Artificial Analysis has recorded, 57%, but a hallucination rate of 86%, against 36% for Opus 4.7 and 50% for Gemini 3.1 Pro Preview, which makes it more likely to answer when it does not know.

Key takeaways · AI-distilled

Per-token pricing doubled from GPT-5.4 to $5/$30 per million input/output tokens, but GPT-5.5 used about 40% fewer output tokens, so the net cost to run the Intelligence Index rose only about 20%.
GPT-5.5 (xhigh) leads Terminal-Bench Hard, GDPval-AA and APEX-Agents-AA. Its GDPval-AA Elo of 1785 is about 30 points ahead of Claude Opus 4.7 (max) and about 470 ahead of Gemini 3.1 Pro Preview.
Effort settings form a cost ladder: GPT-5.5 (low) approximates Claude Opus 4.7 non-reasoning at about half the cost (~$500 vs ~$1,000), though Gemini 3.1 Pro Preview matches GPT-5.5 medium's score for about $900 versus $1,200.
GPT-5.5 (xhigh) posts the highest AA-Omniscience accuracy at 57% but tends to answer when it does not know: its hallucination rate compares with 36% for Opus 4.7 (max) and 50% for Gemini 3.1 Pro Preview.

Terms in this piece · Glossary

token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
hallucination — When a model states something false with full confidence — inventing facts, citations, or APIs that don't exist.

Read the source artificialanalysis.ai