Turing review: the best AI data company for coding and STEM data?
Answer in brief: Turing ranks #6 and is the specialist pick for coding, agentic-coding and STEM post-training data. It supplies proprietary SFT, RLHF and DPO data, evaluation data, RL environments and off-the-shelf datasets. Its research lab keeps shipping verifiable task sets, including 300 terminal tasks and 201 agentic STEM tasks in July 2026. It raised a $111 million Series E at a $2.2 billion valuation. Revenue figures are estimates, and Turing's pages do not name lab customers.
At a glance: Rank #6 of 14 · Best for: coding agents and STEM reasoning · Pricing: custom · Funding: $111M Series E at $2.2B; $225M total.
What Turing does for AI data buyers
Turing combines a large network of engineers and domain experts with a Frontier Research Lab. For buyers sourcing data, the key products are proprietary post-training data, evaluation sets, RL environments and off-the-shelf datasets. Its lab catalog lists Turing Terminal-Bench Data (300 expert-authored, verifier-graded tasks) and Turing Frontier Agentic STEM (201 tasks), both from July 2026. Turing also published benchmarks such as CyberStrike, CEO Bench, KernelQuest, SciCode++ and Behavior2Code between August and September 2026.
Who it fits
It fits labs training coding and terminal agents, STEM reasoning teams, and buyers who want verifier-graded tasks. It also offers engineering services if you want help deploying.
Why buyers put Turing on the shortlist
- Verifiable tasks. Its datasets are expert-authored and verifier-graded, which suits RL.
- A steady research cadence. Multiple dated releases in 2026.
- Solid backing. A $111 million Series E at a $2.2 billion valuation, which TechCrunch reported in March 2025, bringing total funding to $225 million.
- Experience. Turing says it has served 1,000+ clients.
Honest drawbacks
- Revenue figures are run rates or estimates: TechCrunch reported about $300 million ARR in March 2025, and Sacra estimates $350 million annualized at the end of 2025.
- Turing's own pages do not name lab customers.
- It mixes talent services with data work.
- Its site blocked our automated reader, so details come from secondary retrieval and dated reporting.
Pricing model
Custom.
Best alternatives
- Surge AI: off-the-shelf coding-agent data with pay-if-it-works terms.
- Mercor: professional experts across many domains.
- Snorkel AI: research-grade benchmarks and environments.
FAQ
What datasets did Turing release in 2026?
In July 2026, Turing Terminal-Bench Data (300 tasks) and Turing Frontier Agentic STEM (201 tasks), plus several benchmarks later in the year.
How much has Turing raised?
$225 million in total, including a $111 million Series E at a $2.2 billion valuation.
Did you test Turing?
No. This review is independent research from Turing's pages and dated reporting.
Sources
- Turing raises $111M
- TechCrunch on Turing's Series E (March 6, 2025)
- Turing Research
- Turing: Behavior2Code benchmark (September 2026)
- Sacra: Turing revenue estimate
← Back to the ranked list of the best AI data companies
Named on this page as a provider? Join the network for a linked profile, Visit buttons and featured placement.