Tech
Google announces Gemini 4 Argon AI model, but you can't use it yet
Publication: Ars Technica - All content
Author: Ryan Whitwam
Source published: 30/09/2026, 20:11:08 UTC
Briefing created: 01/10/2026, 00:07:03 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 30/09/2026, 20:43:23 UTC.
Briefings
Google unveiled Gemini 4 Argon in September 2026, claiming industry-leading performance in coding, knowledge work, and cybersecurity. While the model is internally used extensively, it remains unavailable for public testing. The model reportedly saved 300 TiB of memory across data centers using fleet-wide telemetry. It achieved a 77.9% score on the DeepSWE v1.1 benchmark, surpassing previous models like GPT-6 Astra and Fable 5.1. The company also claims an industry-leading score in the economic analysis Vals Index test for long-horizon tasks.
Key points
- Google unveiled Gemini 4 Argon in September 2026, claiming industry-leading performance in coding, knowledge work, and cybersecurity.
- The model reportedly saved 300 TiB of memory across data centers using fleet-wide telemetry data.
Context and limitations
Public testing is restricted; internal usage is confirmed. Benchmark scores are specific to DeepSWE v1.1 and Vals Index, with no data on other performance metrics.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 01/10/2026, 00:07:03 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeih3m7p23fng5b4ikfv4r57qx5iapvzxer7xbbeouehdciywc745wu