TL;DR

The short version

Gemini 4 Argon is generally available and scores 53 on Artificial Analysis — back in the frontier cluster with Astra, next to Opus/Sonnet/Wonder Full — at introductory API pricing through 30 November.

On The AI Daily Brief, Nathaniel Whittemore argues the right question is where Argon fits in a multi-model stack. Coding takes look like a harness problem; writing may be the sticky everyday use case.

Based on Nathaniel Whittemore's 14 Oct 2026 episode of The AI Daily Brief, with numbers checked against Artificial Analysis and Google's GA post.

The numbers put Google back

Artificial Analysis scores Argon (high) at 53 on the Intelligence Index — level with GPT-6 Astra, in the same frontier cluster as Opus 5.5, Sonnet 5.5 and Muse Wonder Full. Through 30 November the API is $2 / $10 per million input/output tokens (then $4 / $20). AA's intro cost per Index task is about $1.99.

53AA Intelligence Index (Argon high)
$1.99Cost per AA Index task (intro)
30 NovIntro pricing ends

Artificial Analysis; Google Gemini API pricing

Whereas it might have been before, how good is it with the implication being should I switch, the question is now much more where does it fit?

Nathaniel Whittemore, The AI Daily Brief

Coding reputation is a harness story

Pro subscribers get Argon in the Gemini CLI path. Google's Antigravity coding harness needs Ultra. Testers on the weaker path called it inadequate; Pavel Surin re-ran a bug-hunt in Antigravity and watched the same model jump near Opus 5.5 Max. Dan Shipper's Every team scored 21% on their senior-engineer rewrite — but also found it fast on greenfield apps and shaky on refactors.

Shipper's line captures the gap: a 2026 model trapped in a 2025 harness. When people say Argon can't code, check which door they walked through.

Writing may be the sticky use case

Multimodal launch demos split opinion. The quieter finding for operators: Shipper called Argon his new favourite writing partner after DeepMind hired professional writers to stamp out slop. He keeps ChatGPT and Claude open day-to-day and switches to Gemini 4 to untangle structure and sharpen metaphors.

Agents, banks, and a consumer corpse

Jamie Dimon told JPMorgan's Q3 call the bank will not let arbitrary outside agents in — confirmed to CNBC as a block on credential-sharing access. Shopify went the other way: any personal agent can connect through a supported app. Character.AI's Chapter 11 filing put hard numbers under one consumer bet: ~$10M/month compute vs peak $2.8M/month subscription revenue, with Disney as a $20M stalking-horse for a license to certain IP.

Related Articles

Sources