Model Overview & Capabilities
Gemini 4 Argon is Google’s flagship frontier reasoning model officially announced on October 13, 2026, capturing Rank #4 on our global AI Coding Leaderboard with a verified composite index score of 53.7. The model showcases unprecedented graduate-level scientific and algorithmic depth, delivering an extraordinary 96.5% on GPQA Diamond, 62.0% on SciCode, 55.0% on SWE-bench Verified, and 57.4% on Terminal-Bench 2.0 under standardized test conditions.
Google introduced Gemini 4 Argon with an aggressive introductory rate of $2.00 per million input tokens and $10.00 per million output tokens, giving an effective blended price of $4.00 per million tokens under 3:1 input:output weighting. This directly rivals Claude Sonnet 5.5 ($4.00 blended) and GPT-6 Sol ($4.00 blended) on developer value, achieving an economical $0.0745 per index point while providing sustained generation throughput of 102 tokens per second and a 1.42-second initial response latency.
Operating with a massive 1,048,576-token context window and up to 65,536 output tokens, Gemini 4 Argon is purpose-built for multimodal codebase ingestion, complex architectural diagram synthesis, enterprise repository refactoring, and rigorous mathematical software verification within Google AI Studio and autonomous agent harnesses.
