# Build agents on OpenAI's Codex infrastructure, skip the orchestrator > Source: > Published: 2026-09-11 11:18:21+00:00 OpenAI opened its Agents API in public beta, giving developers the managed sandboxes, long-running sessions and subagent orchestration that power Codex internally through one API call. Read: OpenAI opened its Agents API in public beta, giving developers the managed sandboxes, long-running sessions and subagent orchestration that power Codex internally through one API call. Read: Artificial Analysis found DeepSeek's new V4.1 Flash, with only 8B to 16B active parameters, outscores the 1.6 trillion parameter V4 Pro on its Intelligence Index while costing about four times less per token. Read: Claude Managed Agents now ships a session viewer, reachable through ant beta:sessions connect with an optional web UI, plus an auto mode that checks each tool call against stated intent before running, denying or asking. Read: OpenAI shipped GPT-Live-1 in the API, a full-duplex voice model that listens while it speaks, handles mid-sentence corrections, and hands reasoning and tool calls to a separate backend model for faster, more natural conversations. Read: Cognition shipped SWE-2, a coding model post-trained on Kimi-K3 with tunable reasoning effort, now live in Devin Desktop and CLI, with vendor benchmarks claiming frontier-level FrontierCode scores at a fraction of SWE-1.7's cost. Read: Anthropic published its most detailed threat intelligence report yet, describing sophisticated attempts to misuse Claude for cyberattacks, influence operations, surveillance and bioweapons research, and how each was disrupted. Read: Shopify's engineering team is moving mobile development from React Native back to native Swift and Kotlin, saying coding agents now make writing and maintaining a feature twice, once per platform, no longer the burden it was in 2020.