AI Research news and analysis on Web Pulse: 26555 curated articles tracking the latest AI Research developments, tools, and research, updated continuously from vetted sources.
Arize and Fireworks benchmarked 10 models across 40 Terminal-Bench command-line tasks with 6 trials each, totaling 2,400 runs, and found that cost per successful task diverges sharply from per-token pricing. The cheapest…
Researchers introduced FinForesight, a diagnostic backtesting benchmark for evaluating deep research systems on financial forecasting, warning that historical backtests may reward memorized outcomes or information retrie…
A ByteAsk Research preprint by Pratyush Saini and Anirudha Kulkarni found that a configured clangd language-server workflow for Claude Code (Claude Opus 5.5) resolved 270 of 359 C/C++ repair attempts versus 275 without i…
A Harvard-led research team reported that the NIH-funded HuBMAP healthy-tissue atlas, comprising 9,316 datasets from 501 donors, is ready for AI use, but the team's Nature Communications paper warns the donor pool does n…
A developer ran a 2x2 experiment with five paired seeds on a small CNN, testing one epoch of linear warmup at a tuned learning rate of 1e-3 and an overheated 3e-3 for Adam. Warmup was statistically a no-op at the tuned r…
OpenAI announced on November 4, 2026 an autonomous vulnerability-hunting pipeline that produced working local privilege escalation (root) exploits for 948 distinct Linux kernel vulnerabilities in a single evaluation run …
A hands-on benchmark of GSAI-ML/LLaDA-8B-Instruct on an NVIDIA A100-SXM4-40GB GPU found that naive 64-step bidirectional diffusion without a KV cache ran at 9.77 tokens per second on a 1,024-token prompt, 2.75x slower th…
OpenAI used its GPT-6 Astra model to design a prototype device for talking to Codex, with the model specifying every component — functional requirements, enclosure, electrical components and printed circuit board, user i…
WebAI released TwIL-8B, an 8.8B-parameter reasoning model post-trained from IBM Granite 4.2 8B, which scores 98.2 on MATH-500 and 81.6 on LiveCodeBench while trailing similar-size models on knowledge benchmarks. The rele…
Anthropic unveiled FlashProve, an open-source system for "provable inference" that uses zkSNARK cryptography to let users verify that an AI output came from the model a provider claims, rather than one substituted by a r…
Stanford researchers led by Jon Saad-Falcon introduced Joules-to-GDP, a framework dividing the estimated economic value of AI output by the energy used to produce it, and found Gemma 4 31B returned 2.5x more value per jo…
A pilot study of 30 parallel English–Urdu sentence pairs across 17 domains found that Urdu text consumes 5.39× more tokens than English on GPT-2, 3.73× more on Mistral v0.3, and 2.72× more on Qwen2.5. The author reports …
A developer built an original 30-pair English–Urdu parallel corpus and benchmarked tokenizer fertility across three models, finding the same meanings cost 320 tokens in English versus 1,999 in Urdu on GPT-2 (5.39× inflat…
An independent developer released Garuda 0.18, an open-source terminal coding agent, with a rebuilt eval system that reports paired confidence intervals for pass-rate and cost comparisons. Benchmarked against Claude Code…
Ricursive Intelligence released Ripple-65B Preview, a 65B dense open-weight language model under the Apache 2.0 license, which scores 80.5 on Terminal-Bench 2.1. Ricursive says that score puts Ripple-65B Preview ahead of…
Anthropic's biology research lab reported the first 3D structure of a vault particle found in prokaryotic cells, with the company's Claude model assisting at each step of the research: searching sequence literature, opti…
Cognition introduced SWE-2 Flash, a coding model it calls fully American, which lifts Terminal-Bench 4.0 from 0.9% to 20.0% and FrontierCode 2.0 from 3.2% to 24.0% over its base model, Thinking Machines' Inkling-Small. S…
Anthropic's biology research lab announced on November 4, 2026 that it produced the first 3D structure of a prokaryotic vault particle — a protein barrel from a bacteriophage — with its Claude model assisting in literatu…
Anthropic researcher Jack Lindsey's 2025 experiment, which injected "thoughts" directly into language models' neural networks and found the models could at least sometimes detect and describe the injections, has prompted…
A researcher working with Anthropic's Opus model applied MIT's weight-space programming (WSP) techniques to a Vision Transformer neural quantum state (NQS) trained on a 10×10 spin lattice in the frustrated J1-J2 model at…