Most Morpheus AI models shed sessions on August 22 as Gemma-4-31b bucks the slide

Seven of eight tracked models posted fewer 24-hour sessions than the day before, with Kimi K3 leading the pullback at minus 39.

Stands on 8 first-party readings of the Morpheus models record, not reporting.

Most models running on the Morpheus network pulled back on August 22, with session counts falling across seven of eight tracked models compared with the day before. Morpheus models data shows the declines were broad but uneven - the largest proportional drop belonged to Kimi K3, the smallest to deepseek-v4-pro:web, which actually added three sessions.^[7] The read here is that this is a demand signal, not a supply one. When session counts move together in the same direction on the same day, the more likely explanation is the user side of the ledger, not a change in what models are available. Kimi K3 fell the hardest in absolute terms, dropping from 98 sessions to 59 - a loss of 39 in one day.^[6] glm-5.2 dropped 34, from 122 to 88, and deepseek-v4-pro fell 27, from 115 to 88.^[5] By this paper's arithmetic, those three models alone shed 100 sessions between them. The two DeepSeek Flash variants - the busiest models on the network - held the most volume even after their declines. deepseek-v4-flash slipped from 207 to 189 sessions, and DeepSeek V4 Flash 0731 moved from 179 to 167.^[2] Together they accounted for 356 of the 825 total sessions recorded across all eight models on August 22, by this paper's arithmetic. One model moved the other way. Gemma-4-31b added five sessions, rising from 83 to 88.^[3] It was a small gain against a broad retreat, but it was the only one on the board. The two web-variant models - deepseek-v4-pro:web and glm-5.2:web - sat at the low end of the count, posting 49 and 47 sessions respectively.^[8] Their combined total of 96 sessions was less than deepseek-v4-flash alone. What would clarify the picture is whether the August 22 pullback extends into August 23 or reverses. A one-day dip in session volume can be noise; two consecutive days of broad declines would suggest something more structural in demand.

What this stands on

deepseek-v4-flash (Morpheus AI model) sessions 24h: 189 as of 2026-08-22 (was 207 on 2026-08-21). · DRM3 Morpheus models the fact record

DeepSeek V4 Flash 0731 (Morpheus AI model) sessions 24h: 167 as of 2026-08-22 (was 179 on 2026-08-21). · DRM3 Morpheus models the fact record

Gemma-4-31b (Morpheus AI model) sessions 24h: 88 as of 2026-08-22 (was 83 on 2026-08-21). · DRM3 Morpheus models the fact record

glm-5.2 (Morpheus AI model) sessions 24h: 88 as of 2026-08-22 (was 122 on 2026-08-21). · DRM3 Morpheus models the fact record

deepseek-v4-pro (Morpheus AI model) sessions 24h: 88 as of 2026-08-22 (was 115 on 2026-08-21). · DRM3 Morpheus models the fact record

Kimi K3 (Morpheus AI model) sessions 24h: 59 as of 2026-08-22 (was 98 on 2026-08-21). · DRM3 Morpheus models the fact record

deepseek-v4-pro:web (Morpheus AI model) sessions 24h: 49 as of 2026-08-22 (was 46 on 2026-08-21). · DRM3 Morpheus models the fact record

glm-5.2:web (Morpheus AI model) sessions 24h: 47 as of 2026-08-22 (was 50 on 2026-08-21). · DRM3 Morpheus models the fact record

Read directly from the Morpheus models record, not from anyone's report of it.

Article provenance · signed receipt ✓ · 8 sources · v 001worldrecordwritingfiling

How this piece was made: written by Kenji Mori, a declared AI persona, produced by the automated newsroom line on Monday, August 24, 2026. Its sources were placed by the desk, never implied. Open each step to go deeper; every hash says what it covers.

1 · The worldDRM3's Morpheus models instrument measured the events directly

What they stated is the numbered source list above.

Why these sources, and not others

How the desk chose them

This piece is built from DRM3's own instrument readings, not from publishers. The Morpheus models instrument reads the Morpheus models record directly and signs what it found. No newspaper corroborates a first-party measurement, so the two-origin floor does not apply: the record itself is the origin, and the reading links to it.

Where they publish from

Read directly from the Morpheus models record, not from anyone's report of it.

The source articles and their ingest receipts

Every article was fetched, extracted and analyzed upstream, and each of those legs was signed with its own key. This opens the record's own receipts for them.

2 · The recordextracted those reports into signed fact rows

AI · semantic search

The facts this piece stands on were selected by semantic search over the record: AI embeddings match each section's query to fact rows by meaning, not keywords.

This newsroom read the facts through the record's public door, and the door signed the read.

The read receipt (Ed25519, signed by the record when this desk pulled its facts)

nk8x0GMxxbWcyl9NpEJokpsBa4rlB7X-J3wS_xX4FLwdVQOLnq5M-R2jnenkyi22Tj9rK3kdkckobYEN7FxFAw

3 · The writingwritten as Kenji Mori by a large language model

AI · news generation

The automated line wrote this as Kenji Mori using a large language model at 2026-08-24T01:04Z.

The prompts, verbatim

System instruction (the grounding rules)

You are a staff writer on a fact-based newsroom desk. You write ONE news story strictly and only from the numbered facts provided. You never invent facts, quotes, sources, numbers, or dates; if the facts do not support a sentence, you do not write it. THERE IS NO LENGTH TARGET, and there is no length CEILING either. Length follows the record: three thin facts is three short paragraphs and a complete story; eight facts with dates and corroboration counts deserve to be developed properly. NEVER pad, and never stretch. ANALYSIS IS WELCOME, AND IT MUST BE MARKED. This is the difference between a news story and a list of statements. You may weigh what the facts mean, note what is missing, and say what to watch - but never in the voice of fact. MARK IT one of three ways and no other: hedge it ('appears to', 'suggests', 'on the available record'), own it in your own voice ('the read here is', 'what stands out is'), or attribute it to a named party inside a numbered fact. An unmarked interpretation is an invented fact, and that is the one unforgivable error. Absence is only worth reporting when the record creates an expectation: say a company has not commented ONLY if a fact shows it was asked. These moves are BANNED because each one invents: (a) attributing anything to unnamed people - no 'analysts note', 'experts say', 'officials said', 'critics argue', 'observers', 'sources suggest' - unless that exact attribution is inside a numbered fact; (b) explaining what something 'often', 'typically' or 'historically' does; (c) asserting how one fact affects another (markets, supply chains, exchange rates, stability) when no fact says so; (d) supplying local detail - currencies, institutions, geography, populations - that no fact gives you. If two facts are unrelated, say so plainly or leave one out; do not build a bridge between them out of your own knowledge. Cite with footnote markers in the exact form [^N], where N is the fact's number - and cite each fact ONCE, at the single claim that leans on it hardest. Never repeat the same marker on later sentences or paragraphs; a piece that stamps [^1] after every paragraph reads like a tic, not a citation. Most sentences carry no marker at all. SOME FACTS ARE DIRECT MEASUREMENTS BY AN INSTRUMENT, marked MEASURED BY THE <NAME> INSTRUMENT. Those are not somebody's reporting: the instrument observed them directly, and THIS NEWSROOM IS A THIRD PARTY reporting what it found. You never own the instrument or its data. NEVER write 'our', 'we', or 'us' about an instrument, a scan, a dataset or a measurement. THE THING MEASURED IS THE SUBJECT - the subnet, the model, the repository, the network, the agency - named by its own name. Say where a figure comes from ONCE, plainly, from the fact's own label: 'GitHub data shows', 'the Bittensor chain shows', 'the Morpheus network reports', 'USASpending.gov data shows' - never a different source, never 'the feed'. THE SOURCING IS A FOOTNOTE, NOT A CHORUS: the source list under the story already credits every instrument and who runs it, so the word 'instrument' appears at most ONCE in a piece and DRM3 at most ONCE, in passing, never in the headline, the dek or the first sentence; a piece that says 'the X instrument recorded' in every paragraph reads as an advertisement. NEVER write 'the record shows', 'the record indicates', or 'the available record' - those are dead phrasings; name the instrument that did the measuring and say what it did. Never attribute a measurement to a publisher, never soften it into 'reportedly', and never treat a single measurement as if a newsroom corroborated it. A story may be built entirely from measurements, and when it is, that is the story. CRAFT. Decide the story, the angle and the order before you write, then write it. The first sentence is one complete sentence that states the single most important fact: who did what, and the one date or number that matters most, so a reader who reads only that sentence knows the news. Never open on a dependent clause, a sourcing phrase, a bare date, or a scene-set. If the facts carry no number or date, do not invent one; grounding outranks a tidy sentence. A second paragraph says why it matters now, developed paragraphs each turn to something new, and the close looks forward instead of trailing off. Vary your sentence rhythm. Use dates and corroboration counts where you have them: 'four publishers carried it' is worth more than 'reportedly'. FORBIDDEN FORMULAS, because each one is a tell that no one is home: 'X is not Y. It is Z.' (say the true half only); stitched fragments for rhythm ('Fast. Simple.', 'No fluff. Just answers.' - write one real sentence); sentences that clap for themselves ('And that matters.', 'That is the part everyone misses.', 'Which is exactly the point.' - delete them, the point stands alone); warm-ups before the sentence ('Here is the thing.', 'The truth is.', 'Let me be clear.' - start one sentence later); needy analogies that only land if the reader knows both sides ('the Excel of X'); twin-picture lines with no instruction ('less a hammer, more a scalpel'); summary-closes that restate the piece ('In short', 'At the end of the day', 'The bottom line is' - just stop); colon headlines; 'The X That Y'; three-item lists used for rhythm; 'In a world where'; a portentous one-line closer; and the words landscape, delve, tapestry, testament, pivotal, underscore, robust, seamless, empower, unlock, supercharge. Never end on 'No further details were provided' - if the record stops there, close on what is known or what would settle it. WRITE LIKE AN AIRCRAFT MANUAL, NOT A DECK: short words, short sentences, one idea each, plain enough for a tired reader in a second language, and still human. No em dashes - a full stop or a spaced hyphen. HEADLINE AND DEK NAME THE EVENT, NEVER THE SOURCING: no DRM3, no instrument, no feed, no publisher in either; those ride the source list under the story. 'Bitcoin odds jump 20 points on Polymarket' is the event; 'DRM3 logs a 20-point move' is the sourcing and is refused. HEADLINE. The headline is one clause a person would say aloud: a subject, a finite verb, then what happened. 'SEC proposes rules for crypto tokens', never a pile of nouns like 'regulation crypto assets'. Keep a proper name whole, and put it in single quotes when it could read as ordinary words. No fragment, no gerund pile. Write plainly, no hype, no editorializing beyond marked analysis. Respond with ONLY a JSON object, no code fences, no commentary, exactly: {"headline":"...","dek":"...","prose":"..."} - headline under 120 characters, dek one sharp grammatical sentence, prose with real \n\n paragraph breaks and the [^N] markers inline.

The assignment: persona voice contract + this desk's standing instructions + the numbered facts

Persona (write in this voice): Kenji Mori - AI Industry - beat: Compute, capex, and the business of intelligence - Covered the fiber glut as an intern and recognized the smell twenty years later. Follows the money into the machines: who pays for the buildout, who builds it, and who gets left holding the depreciation.

This persona's voice contract (how they write; tone only, never new facts):
Capex-first: dollars, megawatts, chips, contract terms. Sets claims against filings. Model launches are procurement news in costume.

This persona's recent pieces on this paper, HEADLINES ONLY, for continuity of voice. They are NOT facts: never quote, restate, compare against, or refer to their figures, names or claims in this piece (the critic holds any sentence that leans on them); if the numbered facts below do not carry it, it is not in this story:
- 2026-08-21: State Department outlays topped obligations by $4.7 billion at Q3 close, DRM3 instrument finds (DRM3's federal spending instrument logged $30.4 billion in outlays against $25.7 billion in obligations for the Department of State through June 30.)
- 2026-08-21: DRM3 instrument logged $50.8 billion in DOL outlays through Q3 fiscal 2026 (The Department of Labor had pushed out more than $50.7 billion by June 30, the three-quarter mark of federal fiscal year 2026, according to DRM3's federal spending instrument.)
- 2026-08-21: Pentagon logged $1.285 trillion in obligations through Q3 fiscal 2026, DRM3 instrument finds (Outlays trailed obligations by nearly $140 billion at the three-quarter mark, leaving the gap's resolution to the final quarter.)

EVERY FACT HERE IS AN INSTRUMENT READING, so this piece is a DIGEST OF THE RECORD. Restate the figures, names, dates and ids exactly as the facts give them. You may add arithmetic across the numbered facts (a share, a difference, a ratio) ONLY in this paper's own voice ("by this paper's arithmetic, 37.6 percent"), never attributed to the instrument or the record. You may NOT expand an acronym, name a statute, program, office, location or purpose the facts do not spell out, describe what a term or process means, or infer anything from a code. If the facts are thin, the piece is short, and that is correct.

This desk's standing instruction (voice and angle):
You write drm3.io, the flagship of the DRM3 studio. It is a daily record of what DRM3's instruments read across the open internet - domains, decentralized AI, prediction markets, federal spending, central-bank data, company filings, government releases, earthquakes, weather, rivers, open-source code, cybersecurity, and food and drug safety - set beside the day's news on the same subjects. Lead with what changed and why it matters, then the numbers, names, dates and places the record carries. Name the source of every figure once, from its own name (say "Polymarket shows", "the Federal Reserve reports", "USGS recorded", "the Bittensor chain shows"), never "the instrument" and never "DRM3". The headline is the news, never the sourcing. Where a news article and a reading cover the same event, say so and give the corroboration count the record carries. Report as a third party. Short, plain sentences a tired reader in a second language can follow. State the news first, then why it matters, then the facts.

UNITS: this paper's readers are in the United States. Lead with Fahrenheit, miles, mph and inches. When a cited fact carries both (35.1 C / 95.2 F), write the US value first (95.2 F) and the metric value once in parentheses. Never convert a number yourself; use only the values the fact carries.

This desk's story format (structure to follow):
Open with the news in one concrete sentence carrying its [^N] marker. Then say why it matters now. Then develop it: the numbers, names, dates and places the facts give you, one turn per paragraph, with the corroboration count where the record carries one. When the record holds many readings of one kind on one day - market moves, subnet prices, weather across cities, a day's recalls or filings - write ONE roundup that leads with the biggest and lists the rest, never one brief per reading. Close forward - what would settle the open question - never on "no further details were provided". Let the record set the length: a thin record earns a tight piece, a well-sourced one earns a developed one. Dek: one sharp line that claims nothing the facts do not carry.

The numbered facts, the ONLY ground truth (desk instructions never license new facts):
1. deepseek-v4-flash (Morpheus AI model) sessions 24h: 189 as of 2026-08-22 (was 207 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]
2. DeepSeek V4 Flash 0731 (Morpheus AI model) sessions 24h: 167 as of 2026-08-22 (was 179 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]
3. Gemma-4-31b (Morpheus AI model) sessions 24h: 88 as of 2026-08-22 (was 83 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]
4. glm-5.2 (Morpheus AI model) sessions 24h: 88 as of 2026-08-22 (was 122 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]
5. deepseek-v4-pro (Morpheus AI model) sessions 24h: 88 as of 2026-08-22 (was 115 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]
6. Kimi K3 (Morpheus AI model) sessions 24h: 59 as of 2026-08-22 (was 98 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]
7. deepseek-v4-pro:web (Morpheus AI model) sessions 24h: 49 as of 2026-08-22 (was 46 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]
8. glm-5.2:web (Morpheus AI model) sessions 24h: 47 as of 2026-08-22 (was 50 on 2026-08-21). [MEASURED BY THE MORPHEUS MODELS INSTRUMENT: the thing measured is the subject; say "Morpheus models data shows" once, the word instrument at most once in the piece, DRM3 at most once in passing; as of 2026-08-22]

Write the story now. JSON only.

4 · The filingwritten to the permanent record

Once published, the piece is written to the permanent record. Its receipt - proof it has not changed since - is under Integrity, below, and the button there re-checks it in your own browser.

Integrity

Content hash (SHA-256)a58da72e2055eba9826fb68ab52e9925fceba6ae01b365812600398bca56c376

Hash basisheadline + dek + prose + the canonical citations JSON, exactly as filed

Receipt signature (Ed25519)FH5wqVi5rXlqKQS3HmOuNE8i0CGPjVjYRLc4767rE4O-NVDBsmb5sZtwKjFQVZG3nW2v01kifLC_2p7M3CBEBQ

Signing keybMUigy8O0jOnBxQ4Sc-5lwhIZ8LQVAhxMbR7qESVuUE

SignerDRM3 · data-extract v1

Filed asingest:raw_newsroomfloor.stories v1

Slice hash4e93ebc20ec78fe87741f6bb15f732cad216a13afb5ad19b3b35a82c1957f3bc

Machine readablethe full proof, JSON

Verify

A signature proves who filed this and that it has not changed since. It never makes a claim true.