OpenAI and Anthropic Launch Cheaper AI Models Amid Price War

AI labs shift focus to affordability

OpenAI and Anthropic released competing mid-tier AI models this week, marking a strategic pivot toward cost-conscious business customers rather than raw capability advances. The launches come weeks after both CEOs publicly endorsed slowing development of frontier AI systems.

Anthropic introduced Claude Opus 5.5, which matches the performance of its flagship Fable 5.1 model while cutting operating costs by roughly 40% compared to July's Opus 5 release. The company plans to follow with Sonnet 5.5 and Haiku 5.5 variants in coming weeks, according to details first reported by Fortune.

OpenAI countered with GPT-6 Sol and GPT-6 Luna, derivatives of its flagship GPT-6 Astra model designed for everyday business tasks. The company slashed API pricing by 50% below even its current promotional rates for GPT-5.6, the previous generation.

Pricing comparison reveals aggressive competition

The API cost structures show OpenAI undercutting Anthropic across the board. Claude Opus 5.5 runs $4 per million input tokens and $20 per million output tokens. GPT-6 Sol costs $2 and $10 respectively, while GPT-6 Luna drops to $0.10 and $0.50.

Both companies attribute the reductions to technical improvements—Anthropic cites efficiency gains, while OpenAI points to advances in caching and inference. Each lab emphasized passing savings directly to customers through both lower prices and increased rate limits.

Why it matters

This price war signals a fundamental shift in AI business models. Rather than betting exclusively on increasingly expensive frontier systems, leading labs are competing for mainstream enterprise adoption where cost per task matters more than theoretical capabilities. The dynamic challenges investor assumptions that AI pricing power will grow alongside model performance—a pattern that hasn't held in other technology markets.

CFOs demand ROI as AI budgets tighten

The affordability push reflects changing customer priorities. Randall Hunt, CTO at AI consulting firm Caylent, told Fortune that chief financial officers are applying stricter scrutiny to AI investments after initial deployments failed to deliver promised returns.

"CFOs have seen some of the sticker shock, and they haven't seen some of the gains that were promised in the initial investments," Hunt said. Companies are now calculating cost per task—measuring how long employees take to complete work with AI tools—to justify continued spending.

Ara Kharazian, lead economist at Ramp, characterized the situation as an "all-out price war" that could constrain both companies' long-term profitability. The competition unfolds on two fronts: launching cheaper model variants while simultaneously cutting prices on premium offerings.

Frontier slowdown meets business reality

Neither release represents a frontier model—a major capability leap. Instead, they demonstrate continued progress in making AI more accessible and economical. Anthropic noted that Opus 5.5 underwent its most rigorous safety testing and received the same cybersecurity and biological risk safeguards previously reserved for more advanced systems, honoring CEO Dario Amodei's mid-September call to "pace the frontier."

The real competitive battleground may lie in capturing everyday business users rather than pushing the absolute limits of AI capability. As companies plan budgets for 2027, cost optimization appears to matter more than cutting-edge performance.

These details were first reported by Fortune.