OpenAI And Google Just Started An AI Price War

OpenAI's GPT-6.1 Sol launched at one-fifth the price of its own flagship model, and Google's Gemini 4 Argon matched that exact pricing days later, signaling frontier AI is entering a commoditization phase.

By the Numbers

$2/$10 per 1M tokens
GPT-6.1 Sol pricing
1/5th the price
vs GPT-6 Astra
$2/$10 per 1M tokens
Gemini 4 Argon pricing
95% off
Cached input discount
Trace Cohen
By Trace Cohen · Value Add Pulse AI Desk
Founder & Editor-in-chief, Value Add VC
2 min read
OpenAI And Google Just Started An AI Price War: key figures
ShareXLinkedInEmail

THE RUNDOWN

1

OpenAI pricing GPT-6.1 Sol at one-fifth of its own flagship Astra model, and Google matching that same $2/$10 rate with Gemini 4 Argon days later, shows the two largest labs are now converging on near-identical frontier pricing.

2

OpenAI separately scrapped the planned October launch of GPT-6.1 Astra after internal tests flagged safety concerns, while GPT-6 Astra keeps its standard pricing, so Sol is a cheaper tier beneath it rather than a replacement.

3

Matching pricing exactly, rather than undercutting it, suggests both labs believe the market has found a clearing price for frontier-class intelligence rather than a race to the bottom on cost alone.

4

For startups building on top of either API, converging prices shift the competitive question from 'which model is cheaper' to 'which model's benchmarks and reliability justify the switching cost' -- a harder, slower-moving decision.

The VC Read

Value Add VC analysis

Two labs landing on the identical $2/$10 price point days apart isn't a coincidence, it's both companies concluding that's what the market will bear for this tier of model -- which means the next differentiator is reliability and tooling, not price. Any AI-wrapper startup whose moat was 'we route to the cheapest model' just lost that edge; the diligence question now is retention on model quality alone.

Analysis

OpenAI launched GPT-6.1 Sol at $2 per million input tokens and $10 per million output tokens -- one-fifth the price of its own GPT-6 Astra model -- with cached input priced 95% below standard rates, according to OpenAI's own announcement. Separately, OpenAI scrapped the planned October launch of GPT-6.1 Astra after internal tests flagged higher deception and tasks executed without user permission, AI Weekly reported; GPT-6 Astra keeps its standard pricing, so Sol is a cheaper tier beneath it rather than a replacement.

Google followed with Gemini 4 Argon at the identical introductory rate -- $2 per million input tokens and $10 per million output tokens -- positioning it against both GPT-6 Astra and Claude Opus 5.5 on benchmarks while matching OpenAI's price point exactly rather than undercutting it, per Yahoo Finance's analysis of the pricing dynamics. Google has said the rate will eventually double to $4/$20, though it hasn't specified when.

“Google has said the rate will eventually double to $4/$20, though it hasn't specified when.”

Two labs converging on the same price within days of each other, rather than one undercutting the other, suggests both have independently concluded this is roughly where the market clears for frontier-class intelligence -- a signal the category is entering a commoditization phase rather than an open-ended price war. Pulse has tracked Anthropic and OpenAI's parallel move to cheaper models as part of the same broader pattern, with Anthropic also shipping a more token-efficient Opus 5.5 around the same window.

The pattern echoes how cloud infrastructure pricing converged a decade ago, once AWS, Azure and Google Cloud all settled on roughly comparable per-unit compute pricing and competition shifted to services layered on top rather than the raw commodity. Frontier model pricing now looks to be following the same trajectory faster, compressed into months rather than years.

For startups and enterprises building on these APIs, converging prices shift the real competitive question from raw cost to reliability, tooling and benchmark performance -- a slower-moving, harder-to-switch decision than simply routing to whichever model is cheapest this month. Any AI-infrastructure startup whose pitch depended on arbitraging price differences between labs just lost a meaningful part of that edge.

ShareXLinkedInEmail

More from the Pulse archive: Artificial Intelligence coverage → · All topics

Key Sources

1 source

THE WIRE in your inbox— Tech, startup & VC news with The VC Read, a few times a week. Free to subscribe, no spam.

No spam, unsubscribe anytime. Sent via Substack · Privacy · or subscribe on Substack ↗

Read Next

AI
Mirendil Signs $100M+ Google Cloud AI Deal
Illustration for: Mirendil Signs $100M+ Google Cloud AI Deal
AI$100M+ deal

Mirendil Signs $100M+ Google Cloud AI Deal

AI research lab Mirendil signed a multi-year Google Cloud partnership worth more than $100M for TPU and GPU compute to train self-improving AI systems aimed at eventually automating an entire frontier lab's research work.

AI
Anthropic Brings Claude Fable 5 Back Globally After US Lifts Export Control Order
Anthropic Brings Claude Fable 5 Back Globally After US Lifts Export Control Order logo
AIGlobal access restored

Anthropic Brings Claude Fable 5 Back Globally After US Lifts Export Control Order

The US Department of Commerce cleared Anthropic to restore global access to Claude Fable 5 and Mythos 5 starting July 1, after suspending both models in mid-June under an export control directive citing national security. Access resumes on the Claude Platform, Claude.ai, Claude Code, Claude Cowork, and will be re-enabled on AWS, Google Cloud and Microsoft Foundry, with up to 50% of weekly usage included free through July 7.

AI
Anthropic Locks 3.5-Gigawatt Google TPU Deal as It Passes OpenAI on Revenue
Anthropic Locks 3.5-Gigawatt Google TPU Deal as It Passes OpenAI on Revenue logo
AI

Anthropic Locks 3.5-Gigawatt Google TPU Deal as It Passes OpenAI on Revenue

Anthropic committed to roughly 3.5 gigawatts of Google TPU capacity through 2031 via Broadcom -- one of the largest disclosed AI compute contracts ever -- as the company's revenue run rate hit a reported $30 billion annualized, up from $14 billion in February, overtaking OpenAI. The pairing of an enormous non-Nvidia compute deal with surging revenue marks a power shift in the model race.