Gemini 4 Pro / Flash: benchmarks, hands-on tests and leaks

iruletheworldmo recaps the week, noting Google's Gemini 4 Argon launch, Meta's Muse Wonder flop, and OpenAI reportedly delaying GPT-6.1 Astra over deception concerns.

The poster recaps this week's Muse Wonder and Gemini 4 Argon releases and speculates the next could be GPT-6.1 Astra or Claude Fable 5.5.

After extended testing, the author finds Gemini 4 Argon strong at following instructions and complex tasks, close to Opus 5.5, but it drains usage so fast a few prompts hit the limit.

Marketing tool Migma AI now supports Gemini 4 Argon, generating a three-email launch sequence — announcement, follow-up and reminder — in the user's brand style.

The operator of the Gesshoku Kitan official account says Google's latest AI, Argon, gave solid advice for overhauling its social media strategy, but the usage limit has run out for now.

The author questions claims that Gemini 4 Argon has surpassed Opus, asking whether it reflects real results or hype on the timeline.

Asked to make a 15-second motion graphic video with hyperframes, Gemini 4 Argon surrounded the frame with a wall of text, leaving the author baffled.

TypingMind ran the same prompt through Gemini 4 Argon and GPT-6 Astra and posted a video comparing the two outputs.

In a bridgebench discussion, a user gave a blunt negative take, saying Gemini 4 Argon's performance isn't particularly impressive.

The author argues Google leans on "the end of coding" hype with every model release, and that real coding-task performance matters more than strong benchmarks.

Someone joked that Gemini 4 Argon is really Opus 5.5 underneath and Google is just hoping nobody notices, with an image to match.

The author gave Argon a numbered hand diagram and asked it to locate the keypoints frame by frame in a video, sharing the prompt and results.