AI Wiki — Knowledge Base

I turned Qwen3.8-27B Q2_64 + llama.cpp into a fully TypeSafe AI-compatible Jev-like system. OpenAI API still intact! World’s first Vision-enabled Jev-like model! <10 GB VRAM, 170 ms on an RTX 3090 and ~140 tok/s in chat. 76% vs. 88% Jev-1.13 Acc. on a diverse 22,000-request typed-decision benchmark

Yesterday I released Bonsai-Llama-Jev - a Jev-like Typed Decision inference system for locally operated, sovereign AI. In my new typed-decision-bench, covering more than 22,000 decisions across 175 us

model-releasesr-localllama

23 Sept 2026

Search: “gpt”

SkillGym: Internalizing Human Skills into LLMs for Real-World Problem Solving

TwinCheck: Evidence-Grounded Negative-Twin Verification for Stateful Tool Agents

Validation and Simulation Catch Different Errors: Four Levels of Evaluation for LLM-Generated Circuits

What Looks Like a Capability Limit in Vision-Language Models Is a Readout Limit

Agensh: Scaling Organizational Intelligence to 1,024 Agents

Capable yet Parsimonious: Extracting and Characterizing Hidden Chain-of-Thought in Frontier Models

CliffCompaction: Cost-Efficient Compaction for Long-Horizon Coding Agents

Component Type, Not Reconstruction Error, Predicts Attention Quantization Sensitivity

Do Synthetic Personas Predict Real Audience Response? A Sim-to-Real Study Where a No-Persona Baseline Beats Persona-Based Copy Simulation

Exciting work from NVIDIA. (bookmark it) Interesting to see this approach to turn public Agent Skills into RL environments. Lots of exciteme…

FinFIRST: Benchmarking Search Agents for Financial Information Retrieval, Sourcing and Traceability

FinInteract: Benchmarking Clarification and Intent Integration in Ambiguous Financial Question Answering

FIRE: Failure-Informed Runtime Engineering for Reliable Language-Model Agents

FrontierMath Erdos

Gemini 3.8 TTS Playground

HARMONY: Hierarchical Agentic Reasoning for MONocular Image-to-Scene Synthesis

How SWE-Serve Exposes the Gap Between Local Tests and Live Serving

Pinocchio: Fast Uncertainty Estimates for Black-Box Language Models

Reasoning Topology Matters: A Controlled Study of LLM-Based Cybersecurity Analysis

SPECTRA: Adaptive Execution of Speculative Decoding on a Runtime-Reconfigurable Tiled Architecture

Syndrome, Synergy, and Safety: Structured Reasoning and Knowledge-Driven Alignment for TCM Prescription Generation

Terminal Shrinkage Averaging Reveals a Schedule-Estimator Interaction in LLM Pretraining

Thanks @arena for the recognition! 🏆 Qwen-Image-2.1 is now the #1 open-source model in both the Image Edit and Text-to-Image Arenas. Try it…

Trains but Doesn't Learn: A Post-Training Delivery Benchmark for LLM Agents as Forward-Deployed Engineers

VLM-in-Sandbox: Visual Workspaces for Agentic Visual Reasoning

What is the Difference Between Me and You? Benchmarking the Quality Gap Between Human-Written and AI-Generated Code

A Pinch of SFT, A Dash of RL: When Reinforcement Learning Helps Long-Horizon Advertising Agents

An Empirical Cost Attribution of Context-Compression Gateways in Multi-Turn Coding Agents

As part of our efforts to pace the frontier, we’re committed to supporting independent assessments with deep levels of access across trainin…

Auditing Political Alignment in LLM Assistants: Engagement, Stance, and User Identity

Beyond Accuracy and Surface Fluency: Risk-Sensitive Evaluation of LLMs for Legal Clause Generation

Did a few quick tests and these seem very very legit. 2.6 Pro in particular is doing well in some pretty hard tasks.

Document Retrieval-Aware Chunking (D-RAC): Universal Retrieval-Aware Ingestion of Enterprise Documents via PDF Normalization and Multimodal Markdown Conversion

EmoPose: Vision-Language Model Guided Emotion-Aware Gesture Generation for Humanoid Robots

ESLM: Risk-Averse Selective Language Modeling for Efficient Pretraining

Especially compared by per-task pricing, which is the metric that should matter, I don't think there is anything competitive anywhere in the…

Explain Less, Understand More: Data-Efficient Personalization of Reader-Dependent Jargons

Fairness Beyond Anonymization? Demographic Leakage in German LLM-Generated Resumes

Fathom-Vaidya: Advancing Medical Reasoning with Rubric-Based Rewards

From Tables to Quantified Statements: Evaluating LLM Inference Generation through Executable Verification

Hi Ollama People!

Introducing DigitalOcean Managed Agents: One AI-native stack to power your intelligence

Machine-Interpretable Information: Compiling Documents into Searchable and Readable Protocol States

Math2Visual-X: A Modular Framework for Pedagogically Aligned Lower-Primary Math Visuals Generation

Memento No More: Coaching AI Agents to Master Multiple Tasks via Hints Internalization

Memory vs. Context? Influential Factors of Factual Recall in Language Models

MiMo-V2.6 is 'simply' the best (for now). Despite its simple architecture design it's currently No.1 in the open-weight benchmarks (weighted…