Gemini 4 argon tops 13 of 18 benchmarks google publishes, and its own engineers say the real work…
gemini 4 argon tops 13 of 18 benchmarks google publishes, and its own engineers say the real work is a different story. insiders told the press that day-to-day coding feels weaker than the leaderboard suggests, even as argon leads the vals index at 68.9%.
the gap between what models score and what they ship is the only metric that matters.
Context
Verified date: Google announced Gemini 4 Argon on Sep 30, 2026. Vals AI lists it at #1 of 45 models on the Vals Index with 68.90%, with a 1M context window and token costs of $4.00 in and $20.00 out. Google's blog calls it the leading model on the Vals Index.
On the count of benchmarks: llm-stats and gemini4argon.com both say Google's table has 19 rows, with Argon leading 13, tying one and trailing five.
The 68.9% on the Vals Index matches Vals AI. The post says Argon tops 13 of 18 benchmarks; the two sources that count the table say 13 of 19, so the 18 is off by one in the sources read. The insiders' claim that day-to-day coding feels weaker was not seen in the sources read, so unsupported here, not refuted. The coding rows are where llm-stats says Argon trails. 'The gap between scores and shipping is the only metric' is the author's opinion.
Related work
- Gemini 4 Argon benchmark table ↗19 rows.
- Vals AI model page ↗Index rank.
- llm-stats launch review ↗Oct 7, 2026.
Watch next
- Read Google's comparison table for the five rows where Argon trails.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 1 October 2026 at 09:18 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →