OpenAI’s Decisions API entered public beta on GPT-6 Luna, returning typed probabilities and choices approximately 10 times faster than the Responses API with pricing set at $0.10 per 1 million input tokens and no output charges (MarkTechPost). The API represents a shift toward structured, low-latency inference for classification and routing tasks, distinct from generative text workloads.
AnthropIc expanded Claude Managed Agents to support dynamic workflows, enabling a lead agent to distribute tasks across up to 1,000 sub-agents in parallel (The Decoder). Testing showed a single agent found 27 of 70 hidden bugs in a codebase, while multi-agent orchestration consistently caught 66, demonstrating measurable gains in agent-based problem decomposition. Separately, Microsoft released Microsoft-Decision-1, a 9B decision-scoring model derived from Qwen 3.5 for routing and agent control tasks (MarkTechPost), and Nace AI open-sourced Drex 1.5, a 9B decision model that scores fixed options in a single forward pass and reads up to 128K tokens (MarkTechPost).