Gemini 4 Argon coding test: token-hungry and pricey, $20 plans no longer buy frontier models
bytebot · x · 2026-10-15
A hands-on test of Gemini 4 Argon's coding performance and value:
- Web UI: noticeably token-hungry, eager to add animations, defaults to writing Python to run for you
- Cursor test: a simple task (888 insertions into a mature codebase), followed by 382 lines of review remediation via GPT-6.1 Sol. Root-cause analysis found the test itself encoded a bug, edge cases weren't checked, cross-module severities were missed, and the task was closed before the spec was complete
- Cost: the run burned 38% of Cursor's other-model allowance on the $20 plan using Gemini 4 Argon medium — expensive
- Pricing gripe: Google's Antigravity serves Opus 5.5 and Sonnet 5.5 on the $20 plan but charges extra for Gemini 4 Argon — chat with the frontier, but not code with it
- Bottom line: $20 no longer buys full frontier Gemini or frontier Claude (Fable absent); only ChatGPT still offers frontier Astra at this price
More from coding & agent
- Player-1: RL turns VLM agents into GTA V players with zero game-specific training — yshan2u · 2026-10-15
- Agentic engineering vs AI engineering: one builds software with agents, the other builds agents as the product — Al_Grigor · 2026-10-15
- Cursor + Gemini 4 Argon builds a multiplayer 3D capture-the-flag game in one go — FinanceYF5 · 2026-10-15
- Claude Opus 5.5 levels up codegen but its docs and PR prose are 'unusable' — indragie · 2026-10-15
- Gary Marcus flags AI agent fatigue: managing coding agents burns out engineers — GaryMarcus · 2026-10-15
- The thrill of building isn't gone — you're just not stopping to enjoy it — _jaydeepkarale · 2026-10-15