Gemini 4 Argon matches GPT-6.1 Sol at $2/$10, claims DeepSWE lead

Google DeepMind timed the launch to the last day of September, a day after OpenAI introduced GPT-6.1 Sol. That proximity, and the exact match in pricing, suggests the two labs are watching each other closely. OpenAI had already halved its own pricing from GPT-6 Astra to reach $2/$10, and Google chose to meet that price rather than undercut or exceed it. The result is a commodity floor in the frontier tier, with Anthropic's enterprise-led approach — anchored by Claude Code and quarterly revenue of $11.6 billion — representing a different path.

The benchmark picture is more tentative. Google is leading with DeepSWE v1.1 at 77.9%, but these are vendor-reported figures without independent verification. The caveat is significant, as the industry's track record on self-reported benchmarks is uneven. What is verifiable is the technical specification: a 2 million token context window and a 1 million output token limit, up from 64,000 in prior Gemini models. Google also reports internal benefits, including a 40% improvement in quantum algorithmic optimization, over 300 TiB of memory freed across its data centers, and an 800,000-line rewrite of the Fuchsia Zircon kernel.

The outcome hinges on whether Argon's benchmark lead holds under scrutiny. If it does, Google may have a credible claim to frontier leadership at the same price as OpenAI. If not, the pricing convergence could matter more than the performance claims. Enterprise buyers weighing coding and reasoning workflows are the immediate audience, with the introductory rates set to double to $4/$20 eventually.