OpenAI and Anthropic cut model prices within hours of each other
OpenAI halved prices for GPT-6 Sol and Luna; Anthropic answered with Opus 5.5 at 40% lower cost and, a week later, Sonnet 5.5.
By Tech CEO Daily Staff, Newsroom
· 3 min read
The news
OpenAI released GPT-6 Sol and GPT-6 Luna on September 22, the smaller siblings of its flagship GPT-6 Astra. Sol, aimed at complex work such as coding, costs $2 per million input tokens and $10 per million output tokens; Luna, pitched at high-volume tasks like summarising and extracting information, costs $0.10 and $0.50. OpenAI said both are half the price of their GPT-5.6 predecessors and credited better caching and inference efficiency, according to TechCrunch and TechTarget.
About 90 minutes later, TechCrunch reported, Anthropic launched Claude Opus 5.5. It lists at $4 per million input tokens and $20 per million output tokens, down from $5 and $25 for Opus 5, and cache reads fell to $0.20 from $0.50. Anthropic says the model performs at the level of its larger Fable 5.1 on most work, generates output more than 30% faster and costs 40% less than Opus 5 on typical workloads.
Both companies made competitive claims. OpenAI said its new models beat Anthropic’s Opus and Fable models; Anthropic published benchmarks showing Opus 5.5 ahead of Fable 5.1 on tests including Terminal-Bench 4.0 (66.4% versus 55.8%). None of these results has been independently verified.
On September 28, Anthropic followed with Claude Sonnet 5.5 at $2 input and $10 output, the same list price as GPT-6 Sol. Anthropic kept Sonnet pricing flat but says the new model finishes tasks up to 30% cheaper because it uses fewer tokens.
The numbers
- GPT-6 Sol price (input/output per 1M tokens)
- $2 / $10
- GPT-6 Luna price (input/output per 1M tokens)
- $0.10 / $0.50
- Claude Opus 5.5 price (input/output per 1M tokens)
- $4 / $20
- Claude Sonnet 5.5 price (input/output per 1M tokens)
- $2 / $10
Why CEOs should care
For companies with real AI spend, this is leverage. List prices for top-tier models dropped by a fifth to a half in a single week, and the two leading vendors now price their mid-tier workhorses identically. Finance teams should re-benchmark current contracts and committed-spend deals, and procurement should push for price-protection clauses that pass future cuts through automatically.
Per-token price is only part of the bill. Both companies are now selling on efficiency: fewer tokens per task, faster output, cheaper cached input. The right comparison is cost per completed task on your own workloads. Teams that built a cost-routing layer, sending simple jobs to cheap models and hard ones to expensive models, can capture these savings quickly; teams hard-wired to one model cannot.
The bigger picture
Analysts quoted by TechTarget linked the cuts to pressure from open-source and smaller models, and one called the pace of cuts probably unsustainable. The timing also fits both companies’ push to show fast revenue growth as IPO speculation builds.
What's next
Anthropic says Haiku 5.5 is due in the coming weeks. Watch whether Google responds on Gemini pricing and whether the cuts extend to enterprise seat plans, not just API tokens.
Sources
- PrimaryIntroducing Claude Opus 5.5— Anthropic
- PrimaryIntroducing Claude Sonnet 5.5— Anthropic
- ReportOpenAI launches GPT-6 Sol and Luna, boasting lower cost and fewer mistakes— TechCrunch
- ReportOpenAI takes AI price war to next level with GPT-6 Sol, Luna pricing— TechTarget
- ReportAnthropic releases Opus 5.5 with lower prices and Fable-level performance— TechCrunch
Newsroom
Reporting and analysis from the Tech CEO Daily newsroom. Each story is researched from primary sources — company announcements, regulatory filings and official advisories — and fact-checked before publication.
Spotted an error? Request a correction. Read our editorial standards and AI policy.
The Daily Brief
The technology briefing for people running businesses.
Weekdays at 6 a.m. ET. Free.
