Claude Opus 4.5 uses 76% fewer tokens at medium effort and costs $5/$25 per million tokens, making enterprise AI more accessible?

Claude Opus 4.5 uses 76% fewer tokens at medium effort and costs $5/$25 per million tokens, making enterprise AI more accessible.

Gemini 3 achieves 1501 Elo on LMArena and PhD-level reasoning with a 1M-token context window?

Gemini 3 achieves 1501 Elo on LMArena and PhD-level reasoning with a 1M-token context window.

GPT-5.1, Gemini 3, and Opus 4.5 all released within weeks, accelerating the AI model race.

Models & Releases

Softsnow May 20, 2026

⚡Anthropic, Google, and OpenAI ship major upgrades with coding, reasoning, and cost breakthroughs.

Deep Dive

Key Points

Claude Opus 4.5 uses 76% fewer tokens at medium effort and costs $5/$25 per million tokens, making enterprise AI more accessible.
Gemini 3 achieves 1501 Elo on LMArena and PhD-level reasoning with a 1M-token context window.
GPT-5.1, Gemini 3, and Opus 4.5 all released within weeks, accelerating the AI model race.

Enterprises now have three top-tier, rapidly evolving models — forcing strategic choices on cost, capability, and integration.