Turing

Active.

Sources 1, 2, 3 · checked 2026-11-08

• Turing Frontier Research Lab builds expert-authored benchmarks and datasets to evaluate frontier AI agents (e.g., Terminal-Bench data, Frontier Agentic STEM)
• Model evaluation using structured tests, RL environments, and coding benchmarks such as SWE-bench++, Code Review Bench, VLM-Bench, and private evaluation suites
• Vetted engineers, researchers, PhDs, and domain experts for rubric QA and reasoning evaluation
• RL environments for agent training and evaluation

Sources 1, 2, 3 · checked 2026-11-08

• Frontier labs needing coding, STEM, and agent evaluation suites tied to post-training data
• Teams wanting private benchmark access

Sources 1, 2, 3 · checked 2026-11-08

Contact vendor (get in touch / request benchmark access)

Sources 1, 2, 3 · checked 2026-11-08

• SOC 2 Type II (stated on Turing's Project Lazarus trust page; report under NDA)

Listed as stated by the vendor. Scope can differ by product or subsidiary; request current reports.

Sources 1, 2, 3 · checked 2026-11-08

• No public pricing
• SOC 2 statement is on a product-specific trust page; confirm it covers the evaluation work
• Turing also sells training data; ask how evaluation sets are kept separate

Sources 1, 2, 3 · checked 2026-11-08