|

Google’s Gemini 4 Argon Pulls Level With GPT-6 Astra and Brings a Smaller Bill

Google’s Gemini 4 Argon has matched OpenAI’s GPT-6 Astra on the Artificial Analysis Intelligence Index, scoring 53 factors. At present launch costs, it additionally prices much less to run per check process.

Google DeepMind introduced the mannequin on September 30. For now, solely a group of trusted cyber defenders can use it.

Argon Puts Google Back Among the Top 3 Labs

Artificial Analysis, an unbiased benchmarking agency, stated the rating places Google again among the many prime 3 AI labs. Argon additionally sits 1 level above OpenAI’s GPT-6.1 Sol, which scored 52.

How Google Gemini Argon Performs. Source: X/Artificial Analysis

The mannequin scored 23 factors larger than Gemini 3.1 Pro Preview, Google’s earlier non-Flash mannequin. Earlier in September, Google launched Gemini 3.8 Flash alongside a cybersecurity variant. For comparability, SpaceXAI’s Grok 4.7 scored 46 on the same index in September.

Artificial Analysis credited Argon’s positive factors to fewer hallucinations and stronger agentic abilities. On its AA-Omniscience check, Argon posted a 15% hallucination price, compared with 51% for Astra. However, Argon answered 50% of these questions appropriately, 13 factors beneath Astra.

In agentic work, Argon led AutomationBench-AA at 77.5%, forward of Anthropic’s Claude Sonnet 5.5 at 71.3%. Still, it trailed Sonnet 5.5, Claude Opus 5.5, and Astra on Terminal Bench 4.

Follow us on X to get the most recent information because it occurs

The Smaller Bill Runs on a Launch Discount

Google has set introductory pricing at $2 per million enter tokens and $10 per million output tokens. Those charges double to $4 and $20 as soon as the promotion ends.

At the discounted costs, Artificial Analysis measured Argon’s price at $1.99 per Index process, in opposition to $3.26 for Astra. However, at commonplace charges, the determine rises to $3.98, about 1.2 instances Astra’s price.

The financial savings come from cheaper tokens, since Argon makes use of extra of them. It averaged 62,000 output tokens per process, in contrast with 27,000 for Astra. The mannequin additionally has a 1 million token output restrict.

“To assist Gemini 4 Argon’s capabilities throughout longer, extra complicated use instances, we’re considerably increasing the mannequin’s output token restrict to an industry-leading 1M tokens, up from the earlier 64K tokens,” the group said.

Cyber Defenders Get the First Look at Gemini 4 Argon

Google CEO Sundar Pichai stated groups throughout Google already use it closely, with work starting from coding to quantum computing.

“Importantly, Argon has frontier safeguards, and we’re rolling it out responsibly – it’s with the US gov’t and going to a set of trusted cyber defenders by way of our Fairwind Program right now,” he noted.

Google’s launch publish says Argon can discover, validate, and patch severe software program flaws by itself. For that purpose, trusted defenders and Google’s inner groups will obtain it with out cyber guardrails. 

Paid API prospects and Google AI Ultra subscribers will get entry first when the rollout widens. Google has not stated whether or not the launch low cost will nonetheless apply by then.

The agency additionally stated it’s engaged within the US authorities’s voluntary pre-release access course of. 

Subscribe to our YouTube channel to observe leaders and journalists present knowledgeable insights

The publish (*4*) appeared first on BeInCrypto.

Similar Posts