GPQA Diamond
PulseAugur coverage of GPQA Diamond — every cluster mentioning GPQA Diamond across labs, papers, and developer communities, ranked by signal.
- instance of VIDRAFT 90%
- instance of LiveCodeBench 90%
- instance of Kimi k3 90%
- developed VIDRAFT 90%
- instance of GLM-5.2 90%
- used by Qwen3 8.27B 70%
- instance of MMLU-Pro 70%
- developed by Qwen3 8.27B 70%
- competes with Qwen3 8.27B 70%
- competes with Qwen3.5:9b 70%
- used by LiveCodeBench 70%
- instance of Qwen3 8.27B 70%
17 day(s) with sentiment data
-
TwIL-8B claims top scores but faces scrutiny over benchmarks and licensing
webAI has released TwIL-8B, a fine-tuned version of IBM's Granite 4.2 8B model, which claims superior performance on coding and math benchmarks compared to Qwen3.5-9B. However, a closer examination of the model card rev…
-
Thinking Machines releases 594B Inkling 2 Small, plans open weights
Thinking Machines has released Inkling 2 Small, a 594 billion parameter model that achieves frontier-level performance on benchmarks. The model utilizes a mixture-of-experts architecture, with only 17 billion parameters…
-
VIDRAFT's Darwin-180B-RSI-R3 claims #2 in global math/science reasoning
VIDRAFT has released Darwin-180B-RSI-R3, an open-weight Mixture-of-Experts large language model that utilizes Recursive Self-Improvement (RSI) to enhance its reasoning capabilities without new human-labeled data. The mo…
-
WebAI releases TwIL-8B reasoning model, tops MATH-500 benchmark
WebAI has launched TwIL-8B, an 8.8 billion parameter reasoning model built upon IBM's Granite 4.2 8B. This model excels in mathematics and coding tasks, achieving a score of 98.2 on the MATH-500 benchmark, and is design…
-
Hydrogen-1.0 scientific model challenges GPT-6 Astra with lower costs
Deep Principle has released Hydrogen-1.0, an open-weight scientific model based on GLM-5.3, designed for chemistry, materials science, and life sciences. In evaluations, Hydrogen-1.0 reportedly outperformed GPT-6 Astra …
-
VIDRAFT POCKET LLMs surpass 3.15M downloads, enabling GPU-free AI on consumer hardware
VIDRAFT POCKET has achieved over 3.15 million downloads for its family of GGUF models, enabling users to run large language models on standard consumer hardware without a dedicated GPU. These models leverage sparse Mixt…
-
Deep Principle releases 753B Hydrogen-1.0 for scientific discovery · 6 sources tracked
Deep Principle has released Hydrogen-1.0, a 753 billion parameter open-weight model designed for scientific problem-solving and research assistance. Built upon GLM-5.3 and fine-tuned on scientific data and laboratory re…
-
PRISM algorithm enhances AI reasoning with step-level verification
Researchers have introduced PRISM, a novel inference algorithm designed to enhance the reasoning capabilities of DeepThink systems. PRISM utilizes Process Reward Models (PRMs) to guide resampling, proposal acceptance, a…
-
SenseTime open-sources SenseNova Loop-9B-Preview with looped Transformer design
SenseTime has open-sourced its SenseNova Loop-9B-Preview model, which utilizes a novel Looped Transformer architecture. This design reuses middle layers with shared weights, effectively doubling their computation to inc…
-
Unsloth releases DeepSeek-V4.1 Dynamic V3.1 GGUFs with claimed accuracy gains
Unsloth has released new GGUF versions of the DeepSeek-V4.1-1017 Dynamic V3.1 model, claiming superior accuracy and size efficiency compared to other GGUF formats. Benchmarks across various metrics like AIME25 and LiveC…
-
New NGM module enhances LLMs with training-free memory
Researchers have developed N-gram Memory (NGM), a novel module designed to enhance Large Language Models (LLMs) without requiring additional training. NGM leverages a model's existing pretrained token embeddings as a me…
-
J-Comm framework enables LLM agents to communicate effectively using Jacobian guidance
Researchers have introduced J-Comm, a novel framework designed to enhance latent communication between large language model agents. J-Comm utilizes a training-free Geometric Agreement for Latent Reasoning (GEAR) compone…
-
GLM-5 open-source LLM now runnable locally via Ollama
The GLM-5 large language model, developed by Z.ai, is now available for free local execution via Ollama. This Mixture-of-Experts model boasts 744 billion total parameters with 40 billion active per token, trained on 28.…
-
Hugging Face benchmark leaders lack verification, score gaps minimal
A recent analysis of Hugging Face's official benchmarks revealed that none of the top-ranked models have undergone third-party verification. The study, conducted by VIDRAFT, examined 44 leaderboards and found that while…
-
Hugging Face GPQA leaderboard lacks verified results, rankings statistically uncertain
The Hugging Face GPQA Diamond leaderboard, as of October 22, 2026, displays 115 rows with no verified submissions, indicating a collection of self-reported claims rather than independently checked results. Analysis reve…
-
New CHIMERA dataset boosts LLM reasoning across scientific domains
Researchers have developed CHIMERA, a synthetic dataset of 9,000 samples designed to enhance the reasoning capabilities of large language models (LLMs) across various scientific domains. This dataset addresses challenge…
-
Inception launches Mercury 3 diffusion LLM with 1M context, 1000+ tok/s speed
Inception has launched Mercury 3, a diffusion-based large language model that processes information in parallel rather than sequentially. The company claims it achieves a speed of 1,068 tokens per second on NVIDIA GPUs …
-
VIDRAFT's Darwin-180B-RSI leads open-model benchmarks
VIDRAFT's Darwin-180B-RSI, an open-weight language model with 180 billion parameters, has achieved the top position on both the MMLU-Pro and GPQA Diamond leaderboards, according to independent benchmarks tracked by AIho…
-
Reflection releases 501B parameter open-weight model Beam with 1M context
Reflection has released its new open-weight model, Beam, featuring a sparse mixture-of-experts architecture with 501 billion total parameters and 23 billion active parameters per token. The model boasts a 1 million toke…
-
20+ Jev Decision Model Alternatives Emerge, Challenging TypeSafe's Offering
In response to TypeSafe's Jev decision model, over 20 alternatives have emerged since its launch in September 2026. These alternatives, developed by companies like Cloudflare, OpenAI, and Perplexity, as well as independ…