Gemini 4 Argon matches GPT-6 Astra in intelligence at ~40% lower cost, full benchmarks show

Artificial Analysis published its full evaluation of Google Gemini 4 Argon on October 1: the high-reasoning tier scored 53 on the Intelligence Index, tying GPT-6 Astra and Fable 5.1 and coming in 1 point above GPT-6.1 Sol, ranking 8th among 223 models (the median for comparable models is 26); at the same time, it costs roughly 40% less than GPT-6 Astra, signaling that competition among top-tier models has reached a deadlock. According to @cedricchee, this is Google DeepMind's first proprietary model to surpass the Flash tier in over seven months.

Confirmed

• AutomationBench-AA top score: Gemini 4 Argon scored 77.5%, leading Claude Sonnet 5.5 (max tier, 71.3%) by 6 percentage points (@ArtificialAnlys).
• Intelligence Index of 53: Full benchmark results for the high-reasoning tier were released by Artificial Analysis in two parts, with consistent reports from @haider1, @cedricchee, and others.
• Value for money: @ConsciousWarrior noted that Gemini 4 matches GPT-6 Astra's score at roughly 40% lower cost; @vitaliychiley added that Argon's cost efficiency falls between the GPT 6.1 and GPT-6 Astra generations.
• Leading on enterprise workflows: Some users report it outperforms DeepSWE v1.1 and AutomationBench on enterprise-grade workflows, with coding capability varying by task.
• Output cap: Some users say the output cap increased from 64K when longer reasoning is enabled (reportedly up to one million tokens, pending official confirmation).

Unconfirmed

• The claim that the output cap has been raised to one million tokens appears only in individual user accounts, with no direct backing from official Artificial Analysis charts.

Why it matters

• By matching GPT-6 Astra in intelligence while costing about 40% less, Gemini 4 Argon directly undermines OpenAI's value positioning for high-end models; its lead on enterprise automation and terminal-task benchmarks also shows Google pushing hard into agent/automation scenarios.

2026-10-01 ~ 2026-10-01 · 10 related posts

Full story(4 episodes)→

• Episode 1: Google Officially Launches Gemini 4 Argon(2026-10-01, 38 posts)
• Episode 2: Leaked Benchmarks Show Gemini 4 Argon Topping 12 of 18 Benchmarks(2026-10-01, 2 posts)
• Episode 3: Gemini 4 Argon matches GPT-6 Astra in intelligence at ~40% lower cost, full benchmarks show(2026-10-01, 10 posts)
• Episode 4: Gemini 4 Argon Reportedly Outputs 1M Tokens Per Response(2026-10-01, 3 posts)

• [source] Gemini 4 Argon tops AutomationBench-AA at 77.5%, 6 points ahead of Claude Sonnet 5.5 — ArtificialAnlys · 2026-10-01
• [source] Full benchmark results for Gemini 4 Argon with high reasoning released — ArtificialAnlys · 2026-10-01
• [source] Gemini 4 Argon (High) benchmarks: 53 on AA Intelligence Index, ranks #8 of 223 — ArtificialAnlys · 2026-10-01
• Gemini 4 Argon hits 53 on AA Intelligence Index, matches GPT-6 Astra at 60% of the cost — cedric_chee · 2026-10-01
• Gemini 4 matches GPT 6 Astra on Artificial Analysis benchmark at 40% lower cost — Conscious_Warrior · 2026-10-01
• Gemini 4 Argon scores 53 on Artificial Analysis Intelligence Index, ties GPT-6 Astra — haider1 · 2026-10-01
• Unverified: Gemini 4 Argon reportedly scores 53 on AA Intelligence Index, 1M-token output — cedric_chee · 2026-10-01
• Gemini 4 Argon scores 53 on AA Intelligence Index, cost sits between GPT 6.1 Sol and GPT-6 Astra — vitaliychiley · 2026-10-01
• Gemini 4 Argon Posts 15% Hallucination Rate, Far Below GPT-6 Astra's 45% — i_dg23 · 2026-10-01

1 near-duplicate retellings: Conscious_Warrior