Key facts
- Anthropic's Opus 5.5 offers input and output tokens at $4 and $20 per million, respectively, a 20% reduction from Opus 5.
- Cache reads for Opus 5.5 are $0.20 per million tokens, 60% less than Opus 5.
- Opus 5.5 generates output over 30% faster than Opus 5.
- Anthropic claims Opus 5.5 provides savings closer to 40% for typical workloads due to lower token costs and reduced token usage.
- OpenAI's GPT-6 Sol is priced at $2 per 1 million input tokens and $10 per 1 million output tokens.
- OpenAI's GPT-6 Luna is priced at $0.10 per 1 million input tokens and $0.50 per 1 million output tokens.
- GPT-6 Sol and Luna cost half as much to use as their predecessors, with some benchmarks showing a few percentage points higher capability.
The race for advanced artificial intelligence models has entered a phase focused on cost-effectiveness and efficiency, with major players like Anthropic and OpenAI releasing new models designed to be more affordable for enterprise users. These releases are not primarily about introducing groundbreaking new capabilities but rather about optimizing existing ones to compete with cheaper, open-weight alternatives.
Anthropic's Opus 5.5, an update to its flagship model, aims to be more competitive by reducing costs and increasing speed. The company claims input and output tokens are now priced at $4 and $20 per million, respectively, a 20% decrease from its predecessor, Opus 5. Cache reads, a significant cost factor in many AI tasks, are down 60% to $0.20 per million tokens. Opus 5.5 also generates output over 30% faster than Opus 5, and Anthropic estimates overall savings closer to 40% for typical workloads due to reduced token usage. The model is noted for its capabilities in high-risk areas like cybersecurity and biology, with protections similar to Fable 5.1 in place.
OpenAI has released GPT-6 Sol and Luna, iterative upgrades to its GPT-5.6 family, following its more powerful GPT-6 Astra model. Sol is positioned as an efficient and affordable alternative for daily tasks, while Luna is the fastest and cheapest option. OpenAI states these models were trained using similar methods to GPT-6 Astra and offer modest capability improvements over their predecessors at half the cost. GPT-6 Sol is priced at $2 per 1 million input tokens and $10 per 1 million output tokens, while GPT-6 Luna costs $0.10 for input and $0.50 for output tokens per million.