SELF
AssessTools
A prototype structured executable and linkable format that stores executable contents in SQLite.
Why it's here
Placed in Assess: 4 article(s) of evidence from 3 source(s), led by research-stage coverage, with 0 in the last 30 days. Confidence 59%.
Evidence (4)
- 5Simon Willison·8/24/2026researchExecutable Files Reimagined as SQLite Databases
The post describes SELF, a prototype executable format that stores runnable program data in SQLite and can be executed directly after being marked as executable. The author argues that ELF already behaves like a database in many ways and presents SQLite as a more self-describing, extensible alternative for representing executable and linkable information.
- 8Hacker News·8/19/2026breakthroughOrnith-1.5 advances self-improving open models
Ornith-1.5 is a new family of open models built around an end-to-end self-improvement loop that generates its own tasks, scaffolds, and solution rollouts for reinforcement learning. The release spans 397B MoE, 35B MoE, and 9B dense variants, with reported benchmark results that match or exceed several larger open-source and proprietary models on reasoning, coding, and agentic tasks.
- 8Hacker News·7/24/2026model_releaseFlux 3 Early Access Released
FLUX 3 is a new multimodal foundation model that jointly learns from images, videos, audio, and language in a single architecture. The company says it can generate and transform video with native audio, support text-to-video and image-to-video workflows, and has shown early preference results over competing video models in preliminary evaluations.
- 6Hugging Face Blog·4/15/2026researchVAKRA benchmarks agent reasoning and tool use in enterprise workflows
Hugging Face Blog introduces VAKRA, an executable benchmark for evaluating AI agents in enterprise-like environments with API and document-based tasks. The post also outlines dataset details and observed failure modes, noting that models perform poorly on multi-step reasoning and tool-use workflows.