Anchor 3.0’s purpose-built small language models caught more FINRA violations than every frontier model tested while running up to 34x faster and up to 12x cheaper; generally available to developers today

ZeroDrift, the Enforcement Runtime for AI, announced the general availability of Anchor 3.0, the first family of small language models built to enforce regulatory and company rules on AI-generated communications before they’re sent. The company also published the first benchmark measuring how well AI models enforce FINRA rules on business communications, built on human-written, attorney-labeled data independently produced by Surge AI, the data labeling company trusted by the world’s leading AI labs. Anchor 3.0 caught 95.5% of violations, more than any frontier model tested. It also outperformed GPT-5.6 Sol, the strongest frontier model on the test, on recall, precision and F1 — a measure that balances recall and precision — while running up to 34x faster and up to 12x cheaper. Developers can sign up and start using Anchor 3.0 today.

Financial institutions are already moving AI from experimentation into production. Cambridge research found that 81% of surveyed financial services firms are adopting AI at some level, with traditional financial institutions reporting 45% agentic AI adoption. FINRA’s 2026 oversight report notes that generative AI can implicate supervision, communications, record-keeping, and fair dealing obligations, and regulators have levied billions of dollars in fines for communications failures. Frontier models can catch many violations, but at 12 to 51 seconds per message for the strongest of them, they can only review after the fact. Anchor 3.0 runs the check on every message before it leaves, in about 1.5 seconds on ZeroDrift’s API and under 100 milliseconds self-hosted.

Marketing Technology News: MarTech Interview with Mark Listes, CEO @ Pendulum Intelligence

“Frontier models made it easy to build capable agents. The hard part is running them inside a regulated business, where every message has to follow the rules and the check has to happen every time, before anything goes out,” said Kumesh Aroomoogan, founder and CEO of ZeroDrift. “A frontier model is too slow and too expensive to do that on every message. Anchor is built for exactly this job. Developers keep building with the models they choose, and Anchor makes sure what their agents say is compliant.”

Anchor 3.0 is available in three sizes:

• Anchor 3.0 Mini — 9B parameters (4B active), post-trained from Gemma E4B. The fastest and lowest-cost model in the family, built for the highest-volume AI traffic. Runs ZeroDrift’s pre-built rule packs and flags violations.
• Anchor 3.0 — 9B parameters (4B active), post-trained from Gemma E4B. The flagship model and the one behind the headline results. Runs ZeroDrift’s full library of more than 200 pre-built rules across FINRA, SEC and other regulations, flags the exact lines that violate a rule and rewrites them so the message goes out compliant. Custom company policies are trained into the model through a LoRA adapter.
• Anchor 3.0 Max — 27B parameters, post-trained from Qwen3.8-27B. The most capable model in the family, built to enforce a company’s own policies out of the box with no fine-tuning. Handles long-form content and document attachments with the largest context window in the family.

Marketing Technology News: How MarTech Is Enabling Autonomous Brand Engagement Across Channels?

“ZeroDrift extends Wand’s governance layer to incorporate external regulatory requirements, enabling our agents’ communications to be checked and corrected before they are sent,” said Cristian Felix, Chief AI Architect of Wand AI. “Strong governance is essential for deploying AI at scale, giving financial institutions the confidence and control they need to move from experimentation into production.”