Anthropic's Claude Opus 5 leaks match Fable 5, OpenAI teases next-gen model
Leaked benchmarks show Opus 5 rivals Fable 5, GPT-5 rumors intensify, and Gemini Spark drops
Deep Dive
Anthropic's Claude Opus 5 leaks revealed, with launch delayed and demos showing Fable 5-level performance. OpenAI's next-generation model rumors surface. Major ChatGPT and Codex updates, Google's Gemini Spark rollout, and FLUX 3 also covered.
Key Points
- Claude Opus 5 leaked benchmarks: 89% MMLU, 92% HumanEval, and 2x faster inference vs Opus 4
- Google's Fable 5 leads with 91% MMLU but Opus 5 matches in coding; Gemini Spark now offers 2M context window
- OpenAI's rumored GPT-5 targets 1.5T parameters and agentic workflows; ChatGPT Codex agents now support 100k token context for autonomous coding
Why It Matters
Enterprise AI competition heats up across capability, speed, and safety, reshaping developer tooling and deployment choices.