Skip to content
Fathom Layer

Independent technology index

The technology indexbuilt on verified numbers.

Hardware, software and AI — specifications read from the manufacturer's own pages, checked automatically before anything goes live. Never a paid ranking.

Ask Fathom AI or search the index...
195
Items indexed
50
Reference entries
0
Paid placements
21
Categories
The Fathom signalRecently published
News & LaunchesOpenAI Expands Daybreak Red to AWS GovCloud (US-East)News & LaunchesOpen WebUI Announces v0.13.0 ReleaseNews & LaunchesCloudflare Launches Zero Wire for Critical Vulnerability IntelligenceNews & LaunchesAWS Announces Integration of EDB Postgres AI with Amazon Bedrock AgentCoreNews & LaunchesSupabase Announces Enterprise-Managed Auth for MCP with ClaudeNews & LaunchesOpenAI Increases Bug Bounty Rewards for Security ResearchersNews & LaunchesHugging Face Announces CWE-bench v2 with Enhanced Coverage and TransparencyNews & Launchesn8n Announces Version 2.47.0 with Bug Fixes and Performance ImprovementsNews & LaunchesOpenAI Announces GPT-Rosalind-6 System Card AddendumNews & LaunchesLangChain Announces Release of langchain-anthropic 1.7.8

Fathom / editorial radar Selected stories

What matters now.

Explore all stories
In focus04 stories

Selected by publication date, measured topic attention and available summaries. Attention does not establish a story's importance.

Measured attention

Technology people are watching.

Public signals matched to our catalog. Each source keeps its own metric; attention is not purchase intent.

Explore the measurements →

Hacker News

points on stories that mention it · Source

  1. Mastra

    5,741 · points on stories that mention it

    Window: –

  2. Claude Pro

    ≈ 5,284 · points on stories that mention it

    Window: –

  3. Claude Code

    ≈ 3,380 · points on stories that mention it

    Window: –

Hugging Face

downloads (30 days) · Source

  1. LTX-2.5

    2,493,402 · downloads (30 days)

    Window: –

    Measured as Lightricks/LTX-2.5

Editorial intelligence

Analysis and guides for the decision.

Analysis · Published

iPad mini OLED: A20 Pro chip, arrives Nov 6

The new iPad mini OLED model pairs an 8.5-inch OLED display with Apple's A20 Pro chip and arrives November 6, 2026. Apple announced it on October 26, 2026…

Read
  1. 02Analysis · Published Gemini 4 Carbon: Access, Price and BenchmarksGemini 4 Carbon is a Google model for cyber defense and long tasks, announced 28 Oct 2026; access is limited to cyber defenders, no developer date. Google calls…
  2. Locked compact server enclosure connected to a network security applianceAI-generated illustration03Guide · Published Securing a self-hosted LLM serverRunning a model locally with Ollama, llama.cpp or vLLM is a few commands. Exposing that server safely is the part the quickstarts skip. The failure mode is…
  3. Local workstation beside a monitor with three conceptual software windowsAI-generated illustration04Guide · Published vLLM vs SGLang vs Ollama: which serving engineThe question stopped being "which model?" and became "which engine runs it best on my hardware?". Three names cover most of the decision. Here is how they…

Technology signals

Across the technology index

Electric Vehicles / 2026

Volkswagen ID. Polo

Why it matters

Volkswagen's smallest EV — an electric supermini positioned as the affordable entry to the ID. range, with UK order books now open.

Explore the product record
WLTP rangeup to 454 km52 kWh version · manufacturer-stated
Illustrative image
SOURCE RECORD
Manufacturer source
volkswagen-newsroom.com
Last source read
Specs on record
32
Provenance

The source is the manufacturer domain on this product record. “Last source read” is when that source was fetched; it is not a fresh check of price or availability. Specifications on record are manufacturer statements, not independent testing.

Beyond the lead signal / 02

More from the index.

Different technologies. The same question: what does the evidence actually say?

Illustrative image

Setup Peripherals · elgato.com

Elgato Stream Deck Neo

A compact, intelligent keypad that simplifies daily desktop workflows.

Open product record
9specifications on recordManufacturer record
  1. 02SmartphonesSamsung Galaxy S26 Ultrasamsung.com6specifications on record
  2. 03Local AI HardwareNvidia GeForce RTX 5090 (32GB)nvidia.com32GB GDDR7Memory
  3. 04WearablesOura Ring Gen 3ouraring.com13specifications on record

Images are illustrative. Figures are manufacturer statements or counts from the index; they are not independent test results.

Decision engines

Calculate the decision, not just read about it.

Live estimate / Q4_K_M

Change model size and context. The memory estimate updates from the same calculation used by Hardware Sizer.

Estimated VRAM

23.5GB

Weights
19.4 GB
KV cache
2.6 GB
Overhead
1.5 GB
Parameters
Context

Calculated estimate under stated assumptions, not a measured benchmark or a compatibility guarantee.

Open Hardware Sizer ↗
  1. Total costAPI OptimizerThe break-even point, in months, between paying for a hosted model API and running your own server.
  2. Cost per callToken EconomicsPaste a system prompt and see what processing it costs across models.
  3. CapacityVector Memory EstimatorEstimate vector storage and RAM across Flat, HNSW and IVFFlat indexes. Inspect the planning buffer and overhead assumptions.
  4. Explore all decision tools

Fathom Compare / decision interface

Two products. One clearer decision.

Choose the first product

Then choose a rival

Choose two products in one category to align their records

Compare all data

Or open one already compared

Explore the index

Every category we index.

Intelligence

Compute

Ecosystem & Mobility

Compute

Local AI Hardware

23 items · Open the index →
Illustrative image

In focus / Release year 2025

Nvidia GeForce RTX 5090 (32GB)

The fastest consumer card available, with 32GB of GDDR7 — and the power bill to match.

Open the record

Explore the set 04 records

Curated setup / assembled contextlocal-ai-workstations
Local AI Workstations Over $1600Illustrative image

The system in focus

Local AI Workstations Over $1600

5 recorded components, with the individual product records below.

Explore this setup

Inside this setup 5 recorded components

  1. 01Nvidia GeForce RTX 5090 (32GB)$1999
  2. 02Mac Mini M4 Pro (64GB)$1999
  3. 03GMKtec EVO-X2 (128GB)$3500
  4. 04Mac Studio (M3 Ultra)$3999
  5. 05Mac Studio M2 Ultra$3999

Powered & developed by

Built by Infinity Solutions

The studio behind Fathom Layer designed and operates the machinery under every page: a local AI worker that reads primary sources, a content pipeline with automatic quality gates, and an append-only ledger of dated observations, each tagged with its method and evidence type.

No company, including clients of Infinity Solutions, can pay for placement, scores or conclusions. Rankings are never for sale.

Infinity Solutions

Build log/api/public/proof · read

  1. 512Published entriesmeasured · · fathomlayer.com catalog
  2. 18,440MCP servers looked up in the OSV vulnerability databasemeasured · · registry.modelcontextprotocol.io · osv.dev
  3. 57,351Dated observations recordedmeasured · · Fathom Layer observation ledger

The Fathom standard

Rankings are never for sale.Evidence decides what stays.

03
published items before a category is indexed.
05
structured data points before an item is published.
00
paid placements. Everything else waits.
Read the methodology