Gemini 4 Argon vibe check: strong writer, weak refactorer
- Source
- x.com
- Date
BREAKING: Gemini 4 is OUT and it's SURPRISINGLY GOOD?? I admit I was ready to trash this model qhen I saw the benchmarks but six of us at @every spent the past week testing it across coding, writing, design, and knowledge work, and were pleasantly surprsied. It’s a breakthrough if your company lives in Google’s ecosystem, and a strong option for frontier hoppers as a secondary model for certain tasks. I Here’s our vibe check: My new favorite writing partner. DeepMind hired professional writers with red pens to stamp out AI slop, and it worked. I use it to untangle structure and sharpen metaphors; it also taught me something new tracing in a few of my favorite books. It's got a natural feel for structure and detailed explanations that Astra and Opus lack. Shaky at refactors, fast on greenfield apps. At low effort on a from-scratch build, Gemini 4 built @kieranklaassen a working webcam heart-rate monitor in 35 minutes—matching Claude Opus 5.5’s quality and beating OpenAI’s models on that test. Where it cuts corners is self-verification and existing codebases. In 17 of Kieran’s 38 benchmark runs, it opened and closed its code-review step just 30 milliseconds apart instead of inspecting its changes. And on my Senior Engineer benchmark—refactoring a vibe coded slop codebase—it worked for four hours and 47 minutes and scored just 21% (versus 81% to 85% for GPT-6 Astra, Fable 5.1, Sonnet 5.5, and Opus 5.5). Not great. A 2026 model trapped in a 2025 harness. Antigravity, Google’s desktop agent app, feels six months to a year behind Claude Code and Codex—lacking cloud environments, SSH, a built-in browser, and smart permission checks (leaving you clicking constant pop-ups or using unrestricted "Turbo" mode). Net Result: I keep ChatGPT and Claude open day to day, and switch to Gemini 4 when I need an editorial partner to structure a tough essay, sharpen a metaphor, or close-read a book. Reach for it for live writing collaboration, fast low-effort prototypes, or steerable UI and dashboard design—ideally via Cursor or the API to skip Antigravity. When a model is good enough to force you to use a bad harness, it tells you something. State of Play: Google hadn’t shipped a major flagship upgrade since Gemini 3 Pro last November. For teams whose companies live in Google’s ecosystem, the model you can use inside your approved work tools finally belongs with the frontier pack. DeepMind closed the model gap; now Google needs to bring the app around it into 2026. Read our full vibe check:
@_mohansolo @every Will be excited to try!
Context
Six people at Every spent a week testing Google's Gemini 4 Argon on coding, writing, design and knowledge work, and Dan Shipper's summary of their vibe check splits the model's strengths sharply by task (x.com). For writing, he calls it his new favorite partner, useful for untangling an essay's structure and sharpening metaphors, and credits that to DeepMind hiring professional writers to edit out generic AI prose. On a from-scratch coding task at low effort, it built a working webcam heart-rate monitor in 35 minutes, which Shipper says matched Claude Opus 5.5's quality and beat OpenAI's models on that test.
The weakness is checking its own work and changing code that already exists. In 17 of Kieran Klaassen's 38 runs, the model opened and closed its code-review step 30 milliseconds apart instead of actually reviewing its changes. On Shipper's Senior Engineer benchmark, which asks for a refactor of a messy vibe-coded codebase, it worked for 4 hours and 47 minutes and scored 21%, against 81% to 85% for the leading Anthropic and OpenAI models; the benchmark's scoreboard lists the models it has scored. Shipper also argues that Antigravity, Google's desktop app, trails Claude Code and Codex by six months to a year, lacking cloud environments, SSH, a built-in browser and smart permission checks, so he recommends using the model through Cursor or the API instead.
His overall verdict is that Gemini 4 Argon is a secondary model for people who switch between frontier labs, reached for when writing, quick prototypes or interface design matter, while ChatGPT and Claude stay his daily tools. The full review argues the bigger change is for companies standardized on Google's tools, since Google had not shipped a major upgrade to its flagship model since Gemini 3 Pro the previous November (every.to); the rest of that article is behind a paywall and was not read for this explanation.
Checking sign-in…
Loading comments…