OpenAI ships Astra for Law as a GPT-6 Astra legal configuration
OpenAI launches Astra for Law: GPT-6 Astra plus a U.S. legal search index, with Trusted Access in ChatGPT and Codex before the API.
OpenAI launches Astra for Law: GPT-6 Astra plus a U.S. legal search index, with Trusted Access in ChatGPT and Codex before the API.
OpenAI’s Admin Console walkthrough shows how ChatGPT Work and Codex credits map to classified tasks and merged-code outcomes so admins can argue value, not just spend.
Composio’s Astra bench finds a ~7-point success spread (72.4%–65.5%), 122s–194s per task, and 3–5× tokens when runs fail.
OpenAI set disclosure criteria and process deadlines for disclosing model misalignment and released six reports of unexpected agent and model behavior.
Anthropic is folding Claude chat and Cowork into one assistant, adding Docs and Slides, and rolling the change out first to Pro and Max.
Claude Code 2.1.274 adds a critical-memory warning, CLAUDE_CODE_MCP_STARTUP_WAIT_MS, OpenTelemetry events, and a large MCP and VS Code fix set.
OpenAI is testing labeled, advertiser-backed Sponsored Agents after ChatGPT ads, plus Ads Manager prompts and HubSpot and Shopify integrations.
Command Code’s beta desktop app for Mac, Linux, and Windows bundles the agent runtime with files, Git, terminal, previews, and /design.
Gemini 3.8 Live and Extended Thinking run tools and reasoning in the background so voice agents can keep talking while jobs finish.
Perplexity’s on-device Computer agent reaches Windows RTX PCs with 24GB+ VRAM, keeping local work off the cloud credit meter.
Laravel MCP 1.0 is stable: MCP 2026-07-28, searchable tool catalogs, cache hints, stateless servers, and stricter OAuth with legacy-client fallback.
Claude Code 2.1.273 adds opt-in LLM-gateway hint headers, MCP disconnect alerts, and forking remote-control sessions into local background runs.
Nous Portal adds Hermes Business team invites, shared balance and skills, plus an on-prem or private-cloud Enterprise SKU.
Anthropic opens Salesforce in Claude in beta: 37 sales skills, live CRM context, and governed updates in chat and Cowork.
Claude Code 2.1.271 adds remote fast mode, per-command sandbox hosts, omitClaudeMd, and self-hosted drain telemetry, plus a long reliability pass.
Coder’s Agent Relay now lists Claude Code so cloud sessions can execute in self-hosted workspaces; still closed early access.
Hermes Agent v0.21.3 (v2026.9.14) tags Cloud and Docker so remote-session refresh and state.db writer-handle leaks actually ship.
LangChain explains its Slack paid-media agent: weekly ad and pipeline reports, proposed changes, and a Deep Agents workspace.
Anthropic says Claude-heavy coding lifted CI jobs 25× in six months and forced a rebuild of test-impact analysis.
OpenAI says Perplexity uses GPT-6 Astra to write comms, change software, and watch production, with fewer human check-ins than earlier models.
6–12 Sep 2026
OpenAI’s Cognition case study: Devin plus GPT-6 Astra returns recordings and test reports so engineers review evidence, not every diff.
Anthropic’s Cowork changelog documents a Windows local-file outage after KB5124008 and calmer retries for cloud-bound scheduled tasks and project memory copies.
Claude Code 2.1.270 fixes a 2.1.269 bug that prompted for permission on read-only git commands in long Bash sessions.
OpenAI moved GPT-Rosalind out of research preview for eligible orgs worldwide, still behind trusted access in API, Codex, and ChatGPT Enterprise.