Anthropic's Opus 5.5 Competes With Elorian's Debut Glimpse Model as Specialized Multimodal Alternative

Anthropic released Opus 5.5 in September as a cost-optimized general-purpose frontier model, delivering 40% cost savings, but faces new competition from Elorian's Glimpse, a visual reasoning specialist that launched October 17 at up to 85% lower pricing. Opus 5.5, released September 22, represents Anthropic's refined flagship, achieving performance comparable to Claude Fable 5.1 across diverse tasks while reducing typical workload costs by 40% and accelerating output generation by 30%. The model maintains a 1-million-token context window at standard pricing ($4 per million input tokens, $20 per output), with support for 128K token outputs and prompt caching at tiered rates. According to benchmark evaluations, Opus 5.5 leads on coding benchmarks (Terminal-Bench 4.0: 66.4% vs. GPT-6 Astra's 57.7%) while Astra maintains an advantage on scientific reasoning tasks. Anthropic independently tested the model with Frontier Design and METR; Opus 5.5 achieved 85% reduction in circumvention attempts versus its predecessor and benefits from safeguards for cybersecurity, biology, and model distillation. The model is available across AWS, Google Cloud, and Microsoft Azure.