Claude Code v2.1.269 (Sept 11) added CLAUDE_CODE_WORKFLOW_MAX_CONCURRENT_AGENTS, letting the Workflow tool fan out up to 256 concurrent agents for inference-bound tasks, alongside a VS Code agent map and claude plugin eval for scored plugin testing. It shipped as one line in a changelog the same week Cursor launched a ‘coordinator agent’ product around the same underlying capability.
GitHub’s Project HydraFusion research preview routes each Copilot CLI task across a pool of models — drafting cheaply, escalating only when a quality gate fails. GitHub’s own benchmark table shows it beating Claude Opus 5 on cost in all three tests, but matching or exceeding Opus 5’s quality in only one of them, per VentureBeat’s read of the numbers.
Anthropic’s September 2026 threat intelligence report accuses Moonshot AI of routing almost 300,000 user requests to Claude through 5,380 fraudulent accounts, saving Claude’s reasoning transcripts as training data, and passing the outputs off as its own Kimi model — with more than 23 million such exchanges between May and July. DeepSeek and Xiaomi face similar accusations; China’s Commerce Ministry calls the claims baseless.
OpenAI shipped the Agents API in public beta on September 10, exposing the Codex harness — session management, context compaction, sandboxing, MCP tool wiring — behind a single API. It’s the same managed-agent-loop pitch Anthropic made with Claude Managed Agents in April, arriving five months late and with a rougher enterprise story: US-only data residency and no zero-data-retention option during the beta.
Cognition closed its Series E on September 8 at a $48B valuation on over $2B raised — both figures higher than the $47B/$1B talks Bloomberg reported six days earlier. Annualized revenue hit roughly $900M, up from $492M in May, with a16z and Accel leading a round more than 20 investors deep.
Bloomberg reported September 2 that Cognition, maker of the Devin coding agent, is in talks to raise roughly $1B at a $47B valuation — up from $26B just three months earlier — with annualized revenue above $900M, up from $492M in May. The round hasn’t closed and terms could still change, but the trajectory says more about where investor conviction is heading than any single benchmark score.
Claude Code v2.1.267 (Sept 9) adds a maxEffortLevel setting that lets organizations cap reasoning effort across every provider without hiding the cap from users, plus a long batch of prompt-cache and session-resumption reliability fixes. Verified directly against Anthropic’s own changelog.
Claude Code v2.1.265 (Sept 8) adds folder-based plugin loading via –plugin-dir and fixes a symlink containment bypass in the plugin path check; v2.1.266, shipped the same day, fixes a regression the previous release introduced for gateway/proxy auth. Verified directly against Anthropic’s own changelog.
Anthropic conducted due diligence on Israeli AI-efficiency startup Decart, valued at roughly $6 billion, then walked away without a deal — per Bloomberg, corroborated by multiple outlets, both companies declining comment. The collapse says more about IPO-timing discipline than about Decart.
OpenAI’s September 6 self-report says its research org now runs 3.1 agent-workdays for every human workday, with the median researcher spending $600/day and the 90th percentile $7,000/day on inference. The same report reveals GPT-6 Astra’s Critical cybersecurity classification forced a 59.2% cut to Astra-class GPU allocation for RL training — and OpenAI admits it doesn’t yet know how to safely reach full recursive self-improvement.