Anthropic claude-opus-5 azure/claude-opus-5 Input: text Output: text RuntimeWire coverage Alexandr Wang releases Meta's Muse Spark 1.3 Max after safety testingGitHub launches HydraFusion to make several AI models do one coding jobAnthropic cuts Fable 5.1 cache-read prices 75% for long-running agentsGemini 3.8 Flash goes live as RuntimeWire begins head-to-head testingAlibaba says Qwen3.8-Max leads open-weight models on commerce agent benchmarkRunway says Solaris generates interfaces frame by frame in real timeClaude Opus 5 (Fast) tops Newsroom Reliability v0.2 benchmarkZ.ai GLM 5.3 Flash tops Editorial Craft benchmark at 0.94MiniMax says M3 sent a business email for $0.018When you should choose Conductor Cloud vs just using ReplitNVIDIA's AVO scores 100% on ARC-AGI-3's public setIdler launches AI data lab after a $9M Paradigm-led seed View on OpenRouter. Model data sourced from OpenRouter.