NVIDIA pairs Nemotron with chemistry data and training recipes
NVIDIA pairs Nemotron 3.5 Super VL with LitmusBench v0.2 and chemistry training recipes. Compare the published checkpoints and how the example scores predictions.
Frontier models, comparisons, and how to choose what fits your work.
NVIDIA pairs Nemotron 3.5 Super VL with LitmusBench v0.2 and chemistry training recipes. Compare the published checkpoints and how the example scores predictions.
Sakana AI’s JFinBench covers six Japanese finance exams. The public download includes 264 of 321 questions. See what a full rerun needs.
A UNESCO and UCL study found that climate-answer accuracy fell under repeated challenges. Use a paired question to test for the same failure in…
dspy-wsp 0.2.0 adds Clef and Strands decision-model backends. Their editable components differ, and the reported results also include calibration.
llama.cpp 1.0 removes deprecated APIs, adds FP8 cache support and adds a speech endpoint. Test embedded applications and HTTP clients separately.
MBZUAI’s institute released a 29-billion-parameter diffusion backbone. Its own results show why testing the middle of an action sequence matters.
A proof from Physical Superintelligence links quantum query speedups to input structure. A small example explains the distinction between queries and runtime.
Muse Ember is available through Meta Model API and Muse Code. Its listed input and output rates change above 200,000 input tokens.
Anthropic launched Fable 5.2 and Mythos 5.2 at $8 input and $40 output per million tokens. The models have different access and safeguards.
Gemini 4 Carbon is generally available in the Gemini API. Google lists introductory token rates through November 30 and double those rates afterward.
MIT highlighted WorldParticle on November 9. The research uses a shared particle simulation architecture across six kinds of physical dynamics.
Nous released Arete-27B with Apache 2.0 weights and an 803-prompt evaluation. The reported results apply to the lab’s rubric and test set.