Claude Opus 5.5 and GPT-6: cutting-edge AI gets cheaper. What it means for your company
The last week of September 2026 brought good news for companies that use or are evaluating artificial intelligence: the most capable models on the market are now faster and cheaper. On September 22, Anthropic and OpenAI launched new models just hours apart, and on the 29th OpenAI added another at its developer event.
Claude Opus 5.5: 20% cheaper and faster
Anthropic unveiled Claude Opus 5.5 on September 22. Its API price is USD 4 per million input tokens and USD 20 per million output tokens, compared with USD 5 and USD 25 for its predecessor, Claude Opus 5: 20% less on the standard API rate (Anthropic, 2026). According to the company, it also generates its responses more than 30% faster than its predecessor and is geared toward agentic coding and long-running tasks (Anthropic, 2026). It is available through the API and on Amazon Web Services, Google Cloud and Microsoft Azure.
GPT-6 Sol, Luna and GPT-6.1 Sol: up to 50% less
The same day, OpenAI launched GPT-6 Sol and GPT-6 Luna, faster and more affordable versions based on GPT-6 Astra, its flagship model unveiled earlier in the month. According to OpenAI, their API prices are 50% lower than the promotional prices of GPT-5.6, the previous generation (OpenAI, 2026; TechCrunch, 2026): Sol costs USD 2 for input and USD 10 for output per million tokens, versus USD 4 and USD 20 for GPT-5.6 Sol, and Luna costs USD 0.10 and USD 0.50, versus USD 0.20 and USD 1.20 for GPT-5.6 Luna (VentureBeat, 2026; OpenAI, 2026).
A week later, on September 29, OpenAI introduced GPT-6.1 Sol, an upgrade to Sol for coding and computer use, with standard prices identical to GPT-6 Sol’s, which amount to one fifth of GPT-6 Astra’s: USD 2 versus USD 10 for input and USD 10 versus USD 50 for output per million tokens (OpenAI, 2026; The Next Web, 2026). In addition, Amazon already offers Claude Opus 5.5 and GPT-6 Sol and Luna on its Bedrock service (AWS, 2026), which makes it easier to use them within the cloud a company already has.
What does this mean for your company?
- Revisit the use cases you ruled out on cost: a customer service assistant, document analysis or report automation may pay off today even if they didn’t a month ago.
- Don’t lock yourself into a single provider: prices and capabilities change every few weeks. An architecture that lets you switch models without rebuilding the solution protects your investment.
- Use the right model for each task: the most powerful models for reasoning or coding, and the most affordable ones, such as Luna, for classifying, summarizing or answering frequently asked questions.
- Measure with your own data: the benchmarks each company publishes are not independently verified. What matters is how the model performs with your documents and your processes.
- Protect your information: a lower price doesn’t change the rules on what data can be sent to an external model or how it is protected.
At Wasivia we design generative AI and RAG solutions that choose the right model for each task and measure cost and quality with your data. Book a 30-minute assessment.
References
- Anthropic — Claude Opus 5.5
- OpenAI Developer Community — Announcing GPT-6 Sol and GPT-6 Luna
- TechCrunch — OpenAI launches GPT-6 Sol and Luna (Sep 22, 2026)
- VentureBeat — OpenAI releases GPT-6 Sol and Luna
- OpenAI Developer Community — DevDay 2026 announcements
- The Next Web — GPT-6.1 Sol at a fifth of GPT-6 Astra's prices
- AWS News Blog — Weekly Roundup (Sep 28, 2026)
- Anthropic — Claude API pricing
- OpenAI — API pricing
