Freemium🇺🇸
Independent benchmarking of AI models and API providers on intelligence, speed, latency and price, with image, video and coding arenas.
Scale Labs Leaderboards AI features and pricing (October 2026)
Scale AI's leaderboards with expert-built and partly private evaluations such as Humanity's Last Exam, SWE-Bench Pro and safety tests.
Contact for pricingFree plan
Pricing: Contact for pricing
The leaderboards are freely accessible online, as of September 2026.
What is Scale Labs Leaderboards?
Scale Labs publishes leaderboards that rank AI models on hard, expert-built evaluations across frontier knowledge, agents, coding, multimodal tasks and safety. Some test sets are kept private.
The site lists more than 20 benchmarks, among them Humanity's Last Exam, DrugDiscoveryBench, SWE-Bench Pro, safety benchmarks such as PropensityBench, MASK and Fortress, and multimodal tests such as AudioMultiChallenge and VISTA. Evaluated models come from OpenAI, Anthropic, Google, Meta and open-source developers. A model can only be listed the first time its developer sees the prompts. Scale AI, Inc. in San Francisco runs the leaderboards.
Scale Labs Leaderboards AI features and use cases
- Checks results on unseen questionsPrivate test sets make it harder for models to score well by having seen the questions in training. Teams get a second view next to public benchmarks that may be inflated.
- Compares coding agents on hard tasksSWE-Bench Pro tests software engineering on demanding issues. Engineering managers can see which models handle complex code changes before choosing a coding assistant. The benchmark is one of more than 20 on the site.
- Reviews safety behaviorBenchmarks such as MASK and Fortress examine honesty and safety behavior. Risk and compliance teams can include these results when assessing a model for employee use. The results add an outside view to internal testing.
What sets Scale Labs Leaderboards apart
- Mix of private and open datasets, with a rule that models are listed only on first contact with the prompts.
- Dedicated safety benchmarks next to capability tests.
Where Scale Labs Leaderboards is strongest
United States Available internationally
Scale AI, Inc. is based in San Francisco, according to its privacy policy.
Questions to ask before you buy Scale Labs Leaderboards
Private test sets cannot be inspected, so you rely on Scale's process; combine the results with public benchmarks and your own tests. Scale AI also sells data services to model developers, which is worth knowing when reading its rankings. Most tests are in English.
Scale Labs Leaderboards reviews
No reviews yet. Be the first to share how your team uses it.
Every review shows its verification level. We read each one before it goes live. How we check reviews
Questions about Scale Labs Leaderboards
Are the leaderboards free?
Yes. The leaderboards are freely accessible online. Organizations that want their model evaluated contact Scale Labs.
Why are some test sets private?
Scale combines private datasets to prevent overfitting with open datasets for broad comparison. Questions that are not public are harder to train on, so scores on them are less likely to be inflated.
Which benchmarks are included?
More than 20, including Humanity's Last Exam, SWE-Bench Pro, DrugDiscoveryBench, safety tests like MASK and Fortress, and multimodal tests like VISTA.
Scale Labs Leaderboards update historyno updates yet
No changes recorded for Scale Labs Leaderboards yet.
Changes the vendor made and confirmed user reports, each checked against the vendor's own pages. We review every tool monthly.
Alternatives to Scale Labs Leaderboards
Tools in the same category (Model leaderboards) with the most overlapping markets.
Contact for pricing · Free plan🇺🇸
Public AI arena, formerly LMArena, that ranks models for text, code, vision, search, images and video from anonymous side-by-side votes.
Contact for pricing🇺🇸
Independent evaluator that benchmarks AI models on professional work in law, finance, tax, healthcare, coding and other fields.
Free🇺🇸
Public leaderboard of how often language models add hallucinations when summarizing documents, scored with Vectara's open HHEM model.