1661 articles tagged with Benchmark
DeepSeek V3 (Dec '24) — GPQA 55.7%, MMLU-Pro 75.2%, and 18.4 intelligence points per dollar. That efficiency at this accuracy is rare.https://olud.ai/leaderboard.html#LLM #Benchmar...
Le plus dur n'est pas d'ecrire un agent LLM, c'est de savoir quand il casse.Agent Eval Kit (19 USD) : 90 cas de test prets: 4 suites, runner multi-providers, score par run, seuil d...
Want to run LLMs like TinyLlama or Llama 3 locally on a #RaspberryPi?I published a complete, hands-on benchmark comparing #Ollama Native vs #Docker Compose installations on Raspber...
KoAGE introduces a cutoff-aware benchmark for Korean neologisms with three acquisition and two generation tasks. arXiv cs.CLSource: arXiv cs.CLhttps://arxiv.org/abs/2610.19399#Mach...
Researchers introduced LePhilBench to evaluate legal-philosophical reasoning in large language models. The benchmark comprises two benchmarking tasks and a dataset.Source: arXiv cs...
Hugging Face Space: joelniklaus/harness-leaderboard
Write string: "🎨 Qwen-Image-3.0-Prohttps://olud.ai/media-leaderboard.html#OpenSource #GenerativeAI #StableDiffusion #AI
South Korea’s benchmark Kospi index has dropped by more than 7% as a sell-off of technology shares deepened after U.S. stocks fell on hotter-than-expected inflation data. The Kospi...
📊 Exaone 4.0 1.2B (Non-reasoning) — the actual numbers GPQA: 42.4% MMLU-Pro: 50% Humanity's Last Exam: 5.7% Long Context Reasoning: 0%Measured independently, not self-reported →htt...
🧠 Multiple AI models perform poorly on a newly introduced benchmark. The benchmark appears to present challenges that current approaches have not successfully addressed.💬 Hacker Ne...
📦 Fresh on the leaderboard: Gemini 4 Argon (Google)Context: 1M tokens · $2 in / $10 out per 1M · commercialAll the latest models, tracked hourly:https://olud.ai/latest.html#AI #LLM
Artemis Security stellt Orion-1 vor, ein KI-Modell für Cyberabwehr, das im eigenen Benchmark Claude Opus 5.5 und GPT-6 Sol übertreffen soll. Allgemeine Frontier-Modelle schränken i...
VETNET is an ecosystem of live benchmark datasets and evaluation frameworks that aims to accelerate AI-driven innovation in animal science and veterinary medicine.Source: arXivhttp...
AI trading bot for Polymarket prediction markets: an ML ensemble plus an OpenAI LLM for decisions, a leaderboard copy-trading bot, Rust services, a Postgres ledger, and a Python mo...
In audit, the opinion is the product, and an opinion is only as good as its basis. Being right isn’t...
📦 Fresh open weights on the leaderboard: LFM2.5-24B-A2B (Liquid)Context: 128k tokens · $0.05 in / $0.7 out per 1M · open weightsAll the latest models, tracked hourly:https://olud.a...
Show HN: Parsimony – an open-source benchmark for how much code AI models addhttps://parsimony.lasu.dev/#ai #open-source
Show HN: Parsimony – an open-source benchmark for how much code AI models add
SAN FRANCISCO, Oct. 14, 2026 /PRNewswire/ -- Toptal, the world's largest talent network of highly skilled domain experts, today announced the launch of ExpertCrowd, a platform for ...
AI agents: how can you assess reported speed gains and benchmark scores?https://podcast.aequitus.net/episodes/2026-10-04-saturday-ai-agents-how-can-you-assess-reported-speed-gain-e...
📊 K2-V2 (Medium) scores are out, measured independently, not from its own team. GPQA 59.8%, MMLU-Pro 76.1%, but Humanity's Last Exam drops to 4.4% — the gap between self-reported a...
Benchmarks are essential for evaluating and improving large language models (LLMs). Yet many widely used benchmarks are redundant, making evaluation unnecessarily expensive. Benchm...
With the rapid development of multimodal large language models (MLLMs), evaluating their video understanding capabilities has become increasingly important. Existing video understa...
Video subtitle removal (VSR) is an important video restoration task, yet existing evaluation remains limited by low-resolution datasets, fragmented test sets, and generic video-inp...