Iternal Technologies Launches Ultrabench, the Free AI Benchmark Aggregator Ranking Every Major AI Model on Intelligence, Price and Hardware Fit

News provided by

Oct 01, 2026, 08:50 ET

Ultrabench.ai ranks 309 current AI models against 530 benchmarks, cites a source for every number and gives the data away free

AUSTIN, Texas, Oct. 1, 2026 /PRNewswire/ -- Iternal Technologies (iternal.ai), the Austin-based enterprise AI strategy company, today announced Ultrabench (ultrabench.ai), a free AI benchmark aggregator that ranks every current large language model on one 0-to-100 intelligence index and places each score beside its price, size, memory needs and hardware fit. Since its August debut, Ultrabench has grown from 199 to 309 tracked models.

Ultrabench provides the broadest and most robust set of AI superintelligence benchmark aggregation and ranking for large language models with expertise from the leading AI strategy and technology firm, Iternal Technologies.

Iternal Technologies has delivered thousands of AI implementations and consultations, and used that industry knowledge and insight to produce a highly effective resource for AI implementation teams to use when selecting the right AI models.

"AI benchmarks are a mess, so I fixed it," said John Byron Hanby IV, founder and chief executive of Iternal Technologies. "The data existed, scattered across vendor blogs, arena sites and a dozen single-benchmark leaderboards. There was no aggregate and no simple API. Ultrabench puts the whole model market at your fingertips."

Until now, comparing models meant finding a reasoning score on a launch blog, a coding result on another leaderboard, pricing on an API marketplace and model size, if disclosed, in a model card. Ultrabench puts it all in one row:

Intelligence index: a weighted average of 10 core benchmarks, including GPQA Diamond, SWE-bench Verified, LiveBench, MMLU-Pro and Humanity's Last Exam
Cost: price per million tokens, refreshed daily, plus intelligence per dollar
Hardware fit: parameter count, memory at BF16, INT8 and INT4, and which GPUs can run them
Proof: a source link and trust rating on every score

Ultrabench aggregates published results from more than 40 sources, including Artificial Analysis, Vals.ai, LMArena, Hugging Face, arXiv and ARC Prize. Vendor self-reported scores rank below independent results, and no model earns an index without at least three qualifying core benchmarks. On the Sept. 30 snapshot, Anthropic's Claude Opus 5.5 led at 92.5; the top open-weights model, Moonshot AI's Kimi K3, scored 84.4.

"Within a year, nearly every frontier model will be smarter than a human," Hanby said. "The winning move won't be buying the smartest model. It will be choosing the most cost-efficient model that's smart enough for the job."

All of it is available free at ultrabench.ai, with no API key, account or signup. One call returns the full ranked board, and the index republishes daily.

"Point your AI agent at Ultrabench and ask which model to use," Hanby said. "That's what we built it for."

About Ultrabench

Ultrabench (ultrabench.ai) is an AI benchmark aggregator and LLM leaderboard created by Iternal Technologies. It ranks current LLMs by intelligence, price and hardware fit, with a source on every number and a free, keyless API.

About Iternal Technologies

Iternal Technologies (iternal.ai), founded in 2018 and based in Austin, Texas, builds valuable tools and provides specialized expertise that gives power over your AI transformation, on your terms. It takes organizations from board-ready AI strategy to secure, enterprise deployment with consulting, executive training and software.

Media contact

Ethan Wood, CMO, Iternal Technologies • [email protected] • 888-996-4099

SOURCE Iternal Technologies