OpenAI and Anthropic Slash Prices Amid Snorkel AI Expansion

Avatar photo

ByLisa Grant

September 23, 2026

Major AI labs launched cheaper, more reliable models this week as Snorkel AI secured $350 million to scale the data infrastructure powering the next generation of algorithmic intelligence.

The digital arms race accelerated this week as the industry’s dominant players executed a coordinated pivot toward affordability and infrastructure scaling. OpenAI and Anthropic both unveiled new model iterations designed to lower the barrier to entry for automated systems, while Snorkel AI secured a massive capital infusion to fortify the data pipelines that feed these growing algorithmic engines.

OpenAI introduced GPT-6 Sol and GPT-6 Luna on September 22, targeting professional workflows and high-volume automation. The Sol model is priced at $2 per million input tokens, representing a 50% reduction compared to its predecessors. According to internal evaluations, Sol reportedly cuts factual errors in half, aiming for the reliability of flagship models at a fraction of the cost. The Luna variant offers an even steeper discount at $0.10 per million input tokens, signaling a clear intent to dominate the low-cost API market.

Anthropic responded immediately by launching Claude Opus 5.5. The new model is positioned as a direct competitor to OpenAI’s offerings, boasting a 20% list-price cut from the previous Opus 5. Anthropic claims its new architecture outscores rival models on software-development benchmarks while costing roughly one-third as much to operate. These aggressive price cuts suggest that the initial era of raw performance competition is shifting into a battle for economic ubiquity, where the cost of surveillance and automation is being driven toward zero.

Parallel to these model releases, Snorkel AI closed a $350 million Series E funding round led by Insight Partners and S32, bringing its valuation to $3.5 billion. This surge in capital—up from a $1.3 billion valuation just over a year ago—reflects the growing necessity of specialized data environments. Snorkel has transitioned from simple labeling software to providing finished datasets and reinforcement-learning environments, utilizing tens of thousands of specialists and AI agents to refine the information used to train frontier models.

CEO Alex Ratner attributed the growth to a booming demand for complex training data and simulated environments. The company’s annualized revenue run rate has reportedly jumped from $20 million to over $350 million in a single year. This financial trajectory underscores a critical reality of the modern tech landscape: while the models grab headlines, the true power lies in the control and curation of the data infrastructure.

As these technologies become cheaper and more integrated into the enterprise stack, the oversight of data provenance becomes paramount. The rapid expansion of Snorkel AI, funded by major venture capital interests including Greylock and Wells Fargo, highlights the industrialization of data as a commodity. With OpenAI and Anthropic lowering the cost of deployment, the scale of data consumption is set to reach unprecedented levels, further entrenching the influence of the algorithmic state over daily digital life.

Leave a Reply

Your email address will not be published. Required fields are marked *