Workiva shipped Tie-Out Agent, Benchmarking Agent, and Sustainability Disclosure Agent with traceable audit trails for financial and ESG compliance workflows. Enterprises now have domain-specific agents that reduce manual reconciliation and disclosure gap analysis with built-in governance.

Cursor is developing Sand, a general-purpose workplace agent handling email, spreadsheets, and non-coding tasks, with public iOS beta featuring Remote Control and Automations that trigger from code changes or Slack messages. Developers face expanding competition from IDE-native agents as coding assistants expand into broader office automation.

OpenAI granted selected academic researchers free access to frontier models and science/math/engineering tools via dedicated ChatGPT tier. This signals expansion into research segments while maintaining pricing discipline in API tiers (GPT-5.6 Sol/Terra/Luna unchanged since July 9).

OpenAI's tiered pricing model for GPT-5.6 remains unchanged since general availability on July 9, with caching and batch discounts available. Practitioners can now forecast long-term API costs as the frontier model pricing settles into predictable multi-tier structure.

Claude Mythos Preview currently tops the leaderboard, with GPT-5 family, Claude Opus/Sonnet, Gemini 3 Pro, Grok 4, and DeepSeek V3/R1 dominating frontier tier alongside open-weights. Market consolidation around six-seven labs signals reduced optionality for enterprises seeking alternatives.