Artificial Analysis puts Gemini 4 Argon level with GPT‑6 Astra

Artificial Analysis says Google's Gemini 4 Argon scores 53 on its Intelligence Index, matching GPT-6 Astra at max effort and one point ahead of GPT-6.1 Sol at 52. The firm calls it a 23-point jump from Google's previous flagship. The model is not publicly available yet, and Artificial Analysis says it is rolling out to selected users.

The firm highlights a 15% hallucination rate, which it calls the lowest of any model scoring 45 or more on its index. It puts GPT-6 Astra at 51%. Artificial Analysis reads this as the model more readily admitting what it does not know. That is its interpretation of its own test, not a figure Google has published.

On cost, Artificial Analysis says Argon runs $1.99 per index task at a 50% promotional discount, about 60% of Astra's cost. After the discount ends, it says the figure rises to $3.98, around 1.2 times Astra. Arena, which ran its own head-to-head, said Argon came in 33% cheaper per task than GPT-6.1 Sol and 70% cheaper than Claude Opus 5.5.

Artificial Analysis also ranks Argon first on its AutomationBench at 78%, and lists a 1M-token context window with text, image, video and speech input. Google announced the model on 30 September. Because few people can use it, these results come from the benchmark firms alone, and the promotional price may change what a task costs in practice.