01
OpenAI previews GPT-5.6 Sol Ultrafast mode
OpenAI introduced an Ultrafast processing mode in preview for GPT-5.6 Sol, with TechCrunch reporting up to 14x faster generation and up to roughly 750 output tokens per second for select customers, backed by a Cerebras partnership.
- Validate whether Ultrafast latency and throughput fit your customer-facing or high-volume document workloads before committing to a standard serving profile.
- Treat Ultrafast availability as a capacity-gated feature and require contract language on quotas, regions, and performance SLOs if you plan production use.
- Reassess total cost of ownership for agentic workflows, because higher tokens/sec can reduce end-to-end processing time but may shift bottlenecks to retrieval, tool calls, and downstream systems.
02
CFO says enterprise revenue now exceeds consumer
OpenAI CFO Sarah Friar told investors that enterprise revenue now represents the majority of OpenAI’s revenue and cited an annualized revenue run rate of about $40 billion.
- Expect more enterprise-oriented packaging, support commitments, and procurement-friendly controls as OpenAI optimizes for business retention rather than consumer growth.
- Use the revenue mix shift as leverage in vendor discussions to ask for clearer SLAs, escalation paths, and roadmap visibility for regulated Czech industries.
- Re-check vendor concentration risk, because a faster enterprise pivot can tighten pricing discipline and change discounting behavior mid-contract.
Source — CNBC enterprisebusiness 03
ChatGPT makes GPT-5.6 Luna default for free tiers
Engadget reported that OpenAI moved GPT-5.6 Luna to be the default model for free and “Go” ChatGPT tiers and began rolling out effectively unlimited text chat, with separate limits for images and files.
- Plan for increased shadow AI usage inside your organization and prepare a managed alternative (SSO, DLP, retention) before usage spikes.
- Use Luna as a low-cost baseline model for internal pilots, but set clear guidelines on what must use higher-tier models for accuracy and risk control.
- Update internal training to distinguish “unlimited text” from file/image limits so teams do not design workflows that fail under tier caps.
04
OpenAI appoints Dali Rajic as Chief Revenue Officer
Fortune reported that OpenAI appointed Dali Rajic as Chief Revenue Officer, following the departure of the prior CRO Denise Dresser after less than a year in the role.
- Expect changes in enterprise sales motion that can affect Czech buyers, including partner emphasis, deal governance, and renewal structure.
- Prepare for potential packaging and pricing revisions by aligning internal stakeholders on must-have contract terms before negotiations start.
- Track account support quality during the transition and secure named escalation paths for business-critical deployments.
Source — Fortune go-to-marketleadership 05
Reuters spotlights AI-agent cyber and governance scrutiny
Reuters highlighted ongoing scrutiny over AI-agent security incidents and referenced OpenAI’s focus on cybersecurity risk and governance as governments and oversight bodies increase attention to misuse scenarios.
- Strengthen internal controls around agent deployment by enforcing least-privilege tool access, human approvals for sensitive actions, and auditable logs.
- Include incident reporting and breach-notification clauses in vendor contracts that align with EU and Czech cybersecurity obligations.
- Run tabletop exercises for AI-assisted identity and access misuse, because agent-driven social engineering can bypass traditional perimeter controls.
Source — Reuters securitygovernance