On September 22, 2026, OpenAI and Anthropic launched new artificial intelligence models designed to lower operating costs and API pricing. OpenAI released GPT-6 Sol and Luna with API prices cut by 50%, while Anthropic introduced Claude Opus 5.5 with up to 60% lower costs. Amid ongoing safety discussions, price competition is intensifying as providers aim to ease cost pressures for enterprise clients and expand market share.
Key Takeaways
- OpenAI and Anthropic simultaneously released lower-cost AI models on September 22, 2026.
- OpenAI launched GPT-6 Sol and GPT-6 Luna with API pricing reduced by 50%.
- Anthropic released Claude Opus 5.5, priced at $4 per million input tokens and $20 per million output tokens.
- Claude Opus 5.5 ranked first on the AAII benchmark with 58 points, while GPT-6 Sol placed fifth with 48 points.
- No pricing changes have been confirmed for standard consumer monthly subscription plans.
OpenAI Releases 50% Cheaper GPT-6 Sol and Luna
OpenAI expanded its GPT-6 lineup by introducing GPT-6 Sol and GPT-6 Luna. API rates for both models have been cut by 50% compared to previous GPT-5.6 promotional pricing.
GPT-6 Sol sits directly below the flagship Astra model and is optimized for complex tasks such as coding. GPT-6 Luna is built for high-speed processing of large workloads, including document summarization and data extraction.
Anthropic Claude Opus 5.5 Cuts Costs and Tops Benchmarks
Anthropic launched Claude Opus 5.5 as the initial release in its Claude 5.5 family. Pricing is set at $4 per million input tokens and $20 per million output tokens.
Claude Opus 5.5 is 60% cheaper than the previous top-tier model, Fable 5.1 ($10 input, $50 output), and cuts operating costs by about 40% compared to Opus 5. On the Artificial Analysis Intelligence Index (AAII), it secured the top spot with 58 points, while GPT-6 Sol tied for fifth place with 48 points.
It also outperformed Claude Fable 5.1 across three major coding benchmarks: TerminalBench, FrontierCode, and CursorBench.
Key Specifications and Cost Comparison
Pricing and core characteristics of the newly announced models are detailed below.
| Model | Input (Per 1M Tokens) | Output (Per 1M Tokens) | Key Features |
|---|---|---|---|
| GPT-6 Sol | Standard pricing (50% reduction) | Standard pricing (50% reduction) | Below Astra; optimized for coding and complex tasks |
| GPT-6 Luna | Standard pricing (50% reduction) | Standard pricing (50% reduction) | High-speed processing for bulk summarization and extraction |
| Claude Opus 5.5 | $4 | $20 | #1 on AAII (58 pts); 60% cheaper than Fable 5.1 |
| Claude Fable 5.1 | $10 | $50 | Previous flagship model |
Both OpenAI and Anthropic are focusing on improving cost efficiency per task alongside overall model capability.
Market Drivers and Unconfirmed Details
The price reductions come in response to enterprise concerns over rising AI compute costs and pressure from low-cost open-weight models introduced by Chinese providers such as Alibaba, Moonshot AI, and DeepSeek.
The announcements focus primarily on developer API rates. It remains unconfirmed whether consumer-facing monthly subscriptions, such as $20-per-month tiers, will see price changes. In addition, figures regarding Anthropic’s $965 billion valuation and annualized revenue estimates exceeding $65 billion date back to June–July 2026 estimates and were not newly confirmed alongside this release.