The Gradient Index — 61/100, Warm

Gradient.news' weekly composite score of the AI ecosystem's temperature. Six honest dimensions, grounded data, no hype.

Updated October 19, 2026 · 06:00 UTC Live
The Gradient Index
61 / 100
Warm Above baseline. The ecosystem is shipping.
↑+3 vs last week

Gradient Index 61/100 — Warm

Narrative generation failed this week. Scores were still computed from live collectors, but the editorial commentary fell back to this short notice. The next cron run will regenerate.

12 weeks of the Gradient Index Solid line = weekly total. Dashed = 4-week moving average. Tap or hover any point for details.

Weekly total 4-week moving average
025507510008-0309-1410-19

Score breakdown Six dimensions, each scored honestly against its own source. Missing data stays missing.

◆

Model momentum

Frontier model movement

10 / 20

1 new, 0 quality jumps, 2 price drops, 0 dropped

Source: Diff of the last two leaderboard snapshots: new models, intelligence-index jumps, price drops, deprecations.
▲

Tool momentum

Product layer movement

20 / 20

5 new, 4 momentum risers, 11 weekly moves, 5 dropped

Source: Diff of the last two tools_radar snapshots: new tools, hype_phase upgrades, weekly_note density.
●

News pulse

Volume of AI stories in our feed farm

15 / 15

500 AI articles captured this week versus baseline 120

Source: FreshRSS Gradient label article count this week, aggregating all AI feeds (labs blogs, tech press, ArXiv aggregators, newsletters).
✶

Editorial volume

What we actually wrote about

15 / 15

7 Daily Signals, avg 4.0 topics each

Source: Gradient.news Daily Signal publications count this week, weighted by cluster size (how much each Signal covered).
$

Funding

Real deals this week

1 / 15

1 funding-tagged stories, ~$0M total disclosed

Source: TechCrunch AI tag RSS + optional Crunchbase news. Counted deals in the last 7 days, deal-size weighted where disclosed.
◐

Narrative sentiment

What our editors felt this week

0 / 15

Source: Claude Sonnet 2-pass grounded analysis: reads the 5 latest Daily Signals + model/tool diffs + headlines, returns a score anchored in evidence.
How we build this index

The six dimensions

Model momentum: How much the frontier of AI models actually shifted in the last 7 days. New entrants, quality bumps, price cuts all score. A dead week of incremental releases scores low.

Tool momentum: How much the tooling layer above the models shifted. New products shipping, existing tools gaining momentum, or dying.

News pulse: Volume of AI stories hitting our feed farm this week. Normalized against a rolling baseline — this measures rhythm, not absolute count.

Editorial volume: Our own editorial output as a proxy for density of stories worth writing. When we have more to cover, the ecosystem is hotter.

Funding: Actual funding rounds and M&A activity announced this week. Not a proxy — the real thing.

Narrative sentiment: Editorial reading of the week. Not a raw headline tone classifier — a grounded sentiment based on what the team actually published.

The bands

Scores 0-100 fall into six bands, from Dormant to Euphoric. The band sets the color of the hero and the historical chart line. Bands are static thresholds, not percentiles, so a Warm week means the same thing over time.

The honesty audit

Every dimension is anchored in real data. Model and tool momentum are diffed from the snapshots we already use for the Leaderboard and Tools Radar — no extra scraping. The funding dimension pulls from the TechCrunch AI feed and counts disclosed deal sizes where available. Editorial volume comes from our own publication database. Sentiment is produced by Claude Sonnet reading the last five Daily Signals and the diff data — it is not a free-form vibe check.

What we do not do

We don't invent numbers. If a data source failed, the dimension scores zero and the methodology says so. We don't weight dimensions dynamically — the weights are fixed in the config, transparent, and auditable in the git history of this repository.

— The Gradient editors