Archive edition · Oct 1–15, 2026

Edition — October 1–15, 2026

5 standout picks across 14 categories. Backfilled per our research methodology: candidates sourced from Hacker News and GitHub within this fortnight's window, then checked against each product's own site or repository. Every pick carries a dated evidence link.

AI Compute & GPU Cloud
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Model Serving & Inference
I

Indra

1M-param OCR model, runs on CPU

1M-param OCR model, CPU-fast, near billion-param accuracy.

Indra, from Datalab, is an OCR model that turns PDFs and images into markdown, text or JSON with bounding boxes. At roughly 1 million parameters it runs entirely on CPU — no GPU or PyTorch — yet reaches about 15× Tesseract's speed and scores 66.7 on OmniParseBench versus PaddleOCR's 58.4. The code is Apache-2.0, with a modified OpenRAIL-M license on the model weights for research and personal use.

Why it mattered

A 1M-parameter OCR model that runs on CPU ~15× faster than Tesseract while approaching billion-parameter accuracy — efficient inference from Datalab.

Evidence dated 2026-10-13github ↗official ↗
AI Data Infrastructure
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Development & Orchestration
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Observability & Evaluation
AB

Agent Box

Sandbox that restricts AI agents

OS-isolated, default-deny sandbox for running AI agents.

Why it mattered

Runs AI agents in an OS-isolated, default-deny sandbox with credential injection that keeps secrets out of the agent — guardrails for autonomous execution.

Evidence dated 2026-10-02github ↗
Researched · pending approvalVisit site ↗
AI Coding & Developer Tools
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Agents
A

ARTEX

Autonomous AI penetration-testing system

Autonomous AI system for penetration testing.

ARTEX is a self-hosted, LLM-driven penetration-testing platform where multiple AI agents work together to test targets for vulnerabilities — decomposing goals into executable tasks, running real security tools, and recording evidence that links each finding to captured network traffic. A dual-graph architecture separates what's known about a target (asset graph) from testing progress (exploration graph); a planner agent dispatches work to concurrent worker agents through a recording proxy, producing an auditable chain of reasoning and proof. It's open source under AGPL-3.0 and explicitly restricted to authorized testing — aimed at security professionals doing sanctioned pentests, bug bounties and red-team work.

Why it mattered

An autonomous AI penetration-testing system that drew the fortnight's strongest GitHub momentum for offensive-security agents.

Evidence dated 2026-10-08github ↗
AI Sales
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Voice Agents
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Evals
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Observability
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Healthcare
H

Healthleap

AI that flags at-risk hospital patients

Nightly AI screening of inpatient records for missed conditions.

Healthleap is an AI platform that analyzes hospital inpatient records each night — labs, weights, vital signs and clinician notes — to flag patients at risk of under-diagnosed conditions like malnutrition, delirium, aspiration pneumonia and pressure ulcers, delivering morning risk scores to care teams via a dashboard. It integrates with EHR systems and uses language models to extract clinical indicators from notes, surfacing risk rather than making diagnoses itself. Reported customers include Penn Medicine, Cedars-Sinai, Intermountain, Houston Methodist and Emory Healthcare; in October 2026 it raised $38M ($8M seed co-led by Sequoia and First Round, $30M Series A led by Hummingbird).

Why it mattered

Targets preventable inpatient harm by catching conditions clinicians miss — a concrete, EHR-integrated clinical-AI use case with a dated, TechCrunch-reported raise.

Evidence dated 2026-10-07techcrunch ↗official ↗
AI Biotech & Drug Discovery
No verified pick this edition

No credible in-window candidate met the evidence bar for this category this edition — we flag it rather than force a weak pick.

AI Monetization & Billing Infrastructure
G

Grandet

Cheapest API for any model

Finds the cheapest API/provider per model, via MCP.

Grandet compares LLM API prices across official providers, inference hosts and relays to surface the cheapest option for any given model. It maintains a large database of provider sites, organizes costs by input/output tokens, and separates relay-credit pricing from cash pricing. It exposes an MCP server and a price API, plus a downloadable Skill for Claude, Cursor and similar tools, so an agent can route each call to the lowest-cost provider programmatically.

Why it mattered

Finds the cheapest API/provider for any model across thousands of sources, with an MCP server — cost routing as monetization infrastructure.

Evidence dated 2026-10-05github ↗official ↗