Anthropic and OpenAI cut AI costs, boost coding performance
6.7 Tech Anthropic and OpenAI slashed AI model prices and improved coding, intensifying competition and reducing hallucination rates in the market.
Anthropic has introduced Claude Opus 5.5, its latest flagship AI model designed for coding and knowledge work. This new release offers lower costs and improved performance compared to previous versions. Opus 5.5 is the first in the Claude 5.5 family, positioned as a more affordable and efficient option, costing 20% less than Opus 5 and 60% less than Fable 5.1. Pricing is set at $4 per million input tokens and $20 per million output tokens, with cached reads now $0.20 per million tokens—a 60% reduction. Output generation is also over 30% faster. Opus 5.5 shows significant improvements in agentic coding, scoring 66.4% on Terminal Bench 4.0 and outperforming both Opus 5 and Fable 5.1 on various benchmarks. In response, OpenAI launched GPT-6 Sol and Luna, cutting API costs by 50% compared to GPT-5.6. Sol is priced at $2 per million input tokens and $10 per million output tokens, while Luna is even more affordable at $0.10 and $0.50, respectively. Both companies highlight reduced hallucination rates and enhanced coding performance, intensifying competition in the AI market.
