AI product
An independent platform for evaluating and comparing generative AI models. It publishes third-party assessments and comparative data on model capabilities and performance.
16 uses taken from transcripts — each links to the moment in the video.
Third-party platform identified as a way to run independent model evaluations.
Artificial Analysis provides model comparisons and evaluation benchmarks for measuring AI intelligence, agentic performance, task cost, and efficiency. The video uses its Intelligence Index and other benchmarks to compare Grok 4.6 with competing models.
Source of model benchmark and cost data; Opus 5.5 was also used to extract data from its homepage and build an alternative visualizer.
Compared reasoning-token and output-token counts for different models, including Opus 5 and Opus 5.5.
Benchmark data site whose data Slopalytics was built on, and whose dashboards inspired the UI patterns
Benchmark source for cost-vs-intelligence and tokens-per-task comparisons across models.
Independent benchmarking service the speaker partners with to verify the published performance results; viewers are directed there to check the numbers themselves.
Benchmark site used to compare model prices, cost per intelligence-index task, and the Pareto frontier of cost versus intelligence over time.
Intelligence index and hallucination benchmark used to compare model scores and effective costs
Hosts the private holdout set of the MLCR medical-claims benchmark and publishes model performance results
Independent benchmarking of Kimi K3.1 on Terminal Bench and the Cyber Index
Benchmark site used to pull the intelligence-index task-cost charts showing how the cost of a given intelligence level falls over time
An independent AI benchmarking company that benchmarks the full AI stack — agents, language models, image/video/speech/music models, hardware and inference providers — to help developers compare tradeoffs of intelligence, cost, speed and cache hit rate. Its website hosts 500+ models' results.
Benchmark scores the creator cites for model comparisons
Benchmark site used to compare model scores, task times and speed across models.
Benchmark site used to compare model cost and performance, including comparing subscription cost rather than raw API cost per task
16 in the library.