Llm
- AIAI · 1 min
Researcher uses LLM memory for program analysis in vulnerability research
Jordy Zomer has developed a novel approach to using large language model (LLM) memory for program analysis, particularly in vulnerability research, as detailed on August 28, 2026, on pwning.systems.
29 Aug, 01:28 pm IST - AIAI · 2 min
GLM-5.3 tops LLM leaderboard with best cost-performance ratio
The GLM-5.3 large language model (LLM) achieved the highest score on the Ed-o-meter leaderboard, outperforming models from Anthropic and OpenAI while costing only about one-fifth as much.
23 Aug, 11:28 pm IST - AIAI · 2 min
Researchers reveal method to extract reasoning traces from proprietary LLM APIs
A team of researchers from institutions including the Max Planck Institute and the University of Tübingen demonstrated that proprietary reasoning traces fr…
11 Aug, 11:28 pm IST - AIAI · 2 min
Cactus releases Needle 2, a 14MB agentic LLM for tiny devices
Cactus has launched Needle 2, a 14MB open-source agentic large language model (LLM) designed for phones, wearables, smart home devices, and robots.
11 Aug, 07:28 am IST - AIAI · 2 min
LLM developers face flood of automated pull requests, say experts
Developers working with large language models (LLMs) are experiencing an overwhelming number of automated pull requests (PRs) and issues, according to Jeremy Theocharis, co-founder and CTO at UMH.
17 Jul, 05:28 am IST - AIAI · 2 min
Australian startup Springboards builds LLM Flint to boost response variety
Australian startup Springboards has developed a new large language model (LLM) called Flint, designed to generate a wider variety of responses to open-ended questions.
01 Jul, 09:22 pm IST - AIAI · 2 min
Real-time LLM Inference on Standard GPUs: 3k tokens/s per request
Kog AI has launched a tech preview of its Kog Inference Engine (KIE), achieving real-time large language model (LLM) inference speeds of 3,000 output token…
29 May, 10:01 pm IST - AIAI · 2 min
Various LLM Smells
A writer who began using large language models (LLMs) to enhance their math blog noticed distinctive patterns in AI-generated text over time, according to shvbsle.in.
29 May, 06:33 am IST - AIAI · 2 min
Antigravity 2.0 Tops the OpenSCAD Architectural 3D LLM Benchmark
Antigravity 2.0 has topped the OpenSCAD Architectural 3D LLM Benchmark by successfully generating a detailed parametric CAD model of the Pantheon, according to modelrift.com.
22 May, 08:30 pm IST - AIAI · 2 min
GitHub Trending: rtk-ai/rtk — CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies
The open-source project rtk, hosted on GitHub, offers a command-line interface (CLI) proxy that significantly reduces large language model (LLM) token cons…
20 May, 04:56 am IST - AI · LOCAL INFERENCEAI · LOCAL INFERENCE · 3 min
DeepSeek-V4-Flash: 96GB RAM model revives LLM steering for engineers
DwarfStar 4 project strips llama.cpp to run DeepSeek’s quasi-frontier model locally, enabling real-time activation manipulation without retraining.
17 May, 12:19 am IST - AI · MEMORY EFFICIENCYAI · MEMORY EFFICIENCY · 3 min
Δ-Mem: 1.31× LLM memory boost with 8×8 matrix, Cornell shows
Cornell’s new 8×8 online memory state delivers 31% gain on MemoryAgentBench, no fine-tuning or context expansion needed—India’s edge-AI startups take note.
16 May, 07:20 pm IST - AI INFRASTRUCTUREAI INFRASTRUCTURE · 3 min
GGUF’s single-file LLM format gains traction but lacks ₹0 Cr chat template standard
llama.cpp’s GGUF format simplifies LLM deployment with a single-file approach, but missing chat template and special token standards create interoperability gaps for Indian AI startups.
15 May, 09:51 am IST - GLOBAL-DEVTOOLS · AI IN FINANCEGLOBAL-DEVTOOLS · AI IN FINANCE · 2 min
Kronos AI model hits 24.8k GitHub stars in days—open-source finance LLM
GitHub’s fastest-growing finance AI model offers live demo, fine-tuning tools, and Hugging Face integration—all open-source.
15 May, 03:48 am IST