Loading page…
Company intelligence / Live profile
Track IBM Research's role in the AI economy through Superpower Daily reporting on tools, products, and infrastructure. This live company page updates automatically as new launches, partnerships, product moves, and market signals are published.
Latest coverage
Key developments
vLLM-Hook MIA targets the speed penalty of inspecting a model during generation. IBM reports low overhead in selected tests, but faster intervention is not itself proof of better safety.
ALTK-Evolve turns an agent’s past attempts into prompt instructions. IBM’s benchmark results show gains across two models, but more guidance does not always mean greater reliability.
The Coursera specialization follows one PyTorch model across competing designs. Core labs need no proprietary tools, while hands-on Spyre access is reserved for partnered programs.
The proprietary configuration found a faster design for $299. One search per configuration limits the comparison, and neither price covers the full development workflow.
The workflow measures whether an agent succeeds every time, not just on average. Its promising gains come from IBM’s AppWorld tests, rather than evidence from live deployments.
The release pairs a pretrained system with common data and benchmarks, giving lunar researchers a reusable base for building and comparing models.
Connected coverage
Your company watchlist