---
cursor:
  subagentId: "bc-94d3881a-7089-51d0-892a-4a03c5de6371"
---

# Metrics sweep: inference layer (Chutes SN64, Engy SN53, SayGM SN28, DSperse SN2, Verathos SN96, others)

Access date for every point: 2026-09-24. Tokens-a-day and requests-a-day for Chutes, Engy and SayGM were already done (growth-slides.md, slide 4) and are not repeated. Scratch data is under `/tmp/sweep/inference/` (Chutes users JSON, 16 Engy epoch JSONs, 469 sampled SayGM epoch headers in `saygm_epochs.jsonl`, Engy burn table `engy_burns.json`, Verathos dashboard JSON).

## Series found

### Chutes SN64: Registered users, cumulative

Unit users · cumulative · source type chain · shape **rising** · key `chutes-users-cumulative`

Cumulative by construction. Read from the team's research API (daily rows since 2024-12-03; 639 rows; last row 2026-09-24 = 801,998). Month-end values shown; the last point is 24 Sep 2026. Registrations, not active users; the arXiv trace counted 314,970 users who sent at least one request Apr 2025-Apr 2026, so roughly 40% of registrations were active over a year.

| date | value | verified | source |
|---|---:|---|---|
| 2024-12-31 | 41 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-01-31 | 1,345 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-02-28 | 4,999 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-03-31 | 8,484 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-04-30 | 155,983 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-05-31 | 255,668 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-06-30 | 395,349 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-07-31 | 474,992 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-08-31 | 517,315 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-09-30 | 547,013 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-10-31 | 618,719 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-11-30 | 648,129 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-12-31 | 667,453 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-01-31 | 684,468 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-02-28 | 710,644 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-03-31 | 729,602 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-04-30 | 744,273 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-05-31 | 757,893 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-06-30 | 769,780 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-07-31 | 782,686 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-08-31 | 793,544 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-09-24 | 801,998 | yes | https://chutes.ai/api/research/stats?type=users |

### Chutes SN64: New user registrations per month

Unit users/month · rate · source type chain · shape **falling** · key `chutes-new-users-per-month`

Same API, daily_count summed by month (month only; Sep 2026 is 1-24 Sep, 8,454). Peak 147,499 in Apr 2025 (free-tier era); 2,200-2,850 a week since July 2026. Honest shape: falling since Oct 2025 (71,706), roughly flat at 8-13K a month in 2026.

| date | value | verified | source |
|---|---:|---|---|
| 2024-12-01 | 41 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-01-01 | 1,304 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-02-01 | 3,654 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-03-01 | 3,485 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-04-01 | 147,499 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-05-01 | 99,685 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-06-01 | 139,681 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-07-01 | 79,643 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-08-01 | 42,323 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-09-01 | 29,698 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-10-01 | 71,706 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-11-01 | 29,410 | yes | https://chutes.ai/api/research/stats?type=users |
| 2025-12-01 | 19,324 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-01-01 | 17,015 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-02-01 | 26,176 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-03-01 | 18,958 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-04-01 | 14,671 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-05-01 | 13,620 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-06-01 | 11,887 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-07-01 | 12,906 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-08-01 | 10,858 | yes | https://chutes.ai/api/research/stats?type=users |
| 2026-09-01 | 8,454 | yes | https://chutes.ai/api/research/stats?type=users |

### Chutes SN64: Image-generation (diffusion) requests per day, monthly mean

Unit requests/day · rate · source type chain · shape **peaked** · key `chutes-image-requests-per-day`

Daily series 2025-01-30 to 2026-03-16 (410 days), then the feed stops; treat as discontinued, not zero. Peak month Oct 2025 (6,114 a day; 209,578 diffusion steps a day). Month only. Not in the LLM token series.

| date | value | steps_per_day | verified | source |
|---|---:|---|---|---|
| 2025-01-01 | 2,826 | 49958 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-02-01 | 5,337 | 95231 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-03-01 | 8,706 | 161306 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-04-01 | 1,209 | 14243 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-05-01 | 667 | 10150 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-06-01 | 306 | 5816 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-07-01 | 273 | 5862 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-08-01 | 1,133 | 33277 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-09-01 | 4,092 | 136090 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-10-01 | 6,114 | 209578 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-11-01 | 5,384 | 181965 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2025-12-01 | 5,410 | 168594 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2026-01-01 | 3,779 | 108905 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2026-02-01 | 2,012 | 52814 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |
| 2026-03-01 | 1,258 | 30728 | yes | https://chutes.ai/api/research/stats?type=diffusion (also api.chutes.ai/invocations/stats/diffusion per chute) |

### Engy SN53: Dollars settled to miners per weekly epoch

Unit USD/week · rate · source type chain · shape **rising** · key `engy-usd-settled-per-epoch`

Signed ledger (owner hotkey signature on each epoch JSON); settled_micro summed over miners. Epochs 0-6 were daily (22-29 Jul 2026) and are summed into one 29 Jul point; weekly epochs 7-14 end Wednesdays 08:00 UTC. Rate-card value of miner-served traffic, not audited revenue; first-party-served models excluded. Dip in epochs 12-13 ($14.3K, $12.7K) then a record $34,870 in epoch 14 (16-23 Sep). Epoch 15 (23-30 Sep) in progress: $4,837 after ~1.4 days.

| date | value | epochs | epoch | verified | source |
|---|---:|---|---|---|---|
| 2026-07-29 | 682.44 | 0-6 |  | yes | https://provider.engy.ai/api/subnet/v1/epoch/6 |
| 2026-08-05 | 945.14 |  | 7 | yes | https://provider.engy.ai/api/subnet/v1/epoch/7 |
| 2026-08-12 | 350.28 |  | 8 | yes | https://provider.engy.ai/api/subnet/v1/epoch/8 |
| 2026-08-19 | 3,038.78 |  | 9 | yes | https://provider.engy.ai/api/subnet/v1/epoch/9 |
| 2026-08-26 | 4,466.58 |  | 10 | yes | https://provider.engy.ai/api/subnet/v1/epoch/10 |
| 2026-09-02 | 15,712.24 |  | 11 | yes | https://provider.engy.ai/api/subnet/v1/epoch/11 |
| 2026-09-09 | 14,317.83 |  | 12 | yes | https://provider.engy.ai/api/subnet/v1/epoch/12 |
| 2026-09-16 | 12,726.10 |  | 13 | yes | https://provider.engy.ai/api/subnet/v1/epoch/13 |
| 2026-09-23 | 34,870.18 |  | 14 | yes | https://provider.engy.ai/api/subnet/v1/epoch/14 |

### Engy SN53: Requests served per weekly epoch

Unit requests/week · rate · source type chain · shape **rising** · key `engy-requests-per-epoch`

Same ledger; requests summed over miners. 29 Jul point is epochs 0-6 summed. Record 7.03M in epoch 14 (1.0M a day).

| date | value | epochs | epoch | verified | source |
|---|---:|---|---|---|---|
| 2026-07-29 | 1,865,752 | 0-6 |  | yes | https://provider.engy.ai/api/subnet/v1/epoch/6 |
| 2026-08-05 | 1,909,616 |  | 7 | yes | https://provider.engy.ai/api/subnet/v1/epoch/7 |
| 2026-08-12 | 449,571 |  | 8 | yes | https://provider.engy.ai/api/subnet/v1/epoch/8 |
| 2026-08-19 | 1,685,453 |  | 9 | yes | https://provider.engy.ai/api/subnet/v1/epoch/9 |
| 2026-08-26 | 1,898,169 |  | 10 | yes | https://provider.engy.ai/api/subnet/v1/epoch/10 |
| 2026-09-02 | 4,263,117 |  | 11 | yes | https://provider.engy.ai/api/subnet/v1/epoch/11 |
| 2026-09-09 | 3,566,420 |  | 12 | yes | https://provider.engy.ai/api/subnet/v1/epoch/12 |
| 2026-09-16 | 3,805,890 |  | 13 | yes | https://provider.engy.ai/api/subnet/v1/epoch/13 |
| 2026-09-23 | 7,031,657 |  | 14 | yes | https://provider.engy.ai/api/subnet/v1/epoch/14 |

### Engy SN53: Miners in the settlement ledger per epoch

Unit miners · rate · source type chain · shape **peaked** · key `engy-miners-paid-per-epoch`

Count of miner rows in each epoch JSON. Peaked at 199 (epoch 10, 19-26 Aug), 75-82 since epoch 12 after the qwen3.6-35b fleet thinned; tokens and dollars rose while miner count fell (consolidation onto Kimi K3 / DeepSeek / GLM miners).

| date | value | epochs | epoch | verified | source |
|---|---:|---|---|---|---|
| 2026-07-29 | 45 | 0-6 max |  | yes | https://provider.engy.ai/api/subnet/v1/epoch/6 |
| 2026-08-05 | 105 |  | 7 | yes | https://provider.engy.ai/api/subnet/v1/epoch/7 |
| 2026-08-12 | 154 |  | 8 | yes | https://provider.engy.ai/api/subnet/v1/epoch/8 |
| 2026-08-19 | 171 |  | 9 | yes | https://provider.engy.ai/api/subnet/v1/epoch/9 |
| 2026-08-26 | 199 |  | 10 | yes | https://provider.engy.ai/api/subnet/v1/epoch/10 |
| 2026-09-02 | 163 |  | 11 | yes | https://provider.engy.ai/api/subnet/v1/epoch/11 |
| 2026-09-09 | 77 |  | 12 | yes | https://provider.engy.ai/api/subnet/v1/epoch/12 |
| 2026-09-16 | 75 |  | 13 | yes | https://provider.engy.ai/api/subnet/v1/epoch/13 |
| 2026-09-23 | 82 |  | 14 | yes | https://provider.engy.ai/api/subnet/v1/epoch/14 |

### Engy SN53: Distinct models with paid traffic per epoch

Unit models · rate · source type chain · shape **rising** · key `engy-models-served-per-epoch`

Distinct `model` values among miner rows. 1 model (qwen3.6-35b-a3b) through epoch 8; 5 from epoch 9 (12-19 Aug); 8 in epochs 13-14 (adds glm-5.3, deepseek-v4.1-flash). Flat at 8 for the last two epochs.

| date | value | epochs | epoch | verified | source |
|---|---:|---|---|---|---|
| 2026-07-29 | 1 | 0-6 |  | yes | https://provider.engy.ai/api/subnet/v1/epoch/6 |
| 2026-08-05 | 1 |  | 7 | yes | https://provider.engy.ai/api/subnet/v1/epoch/7 |
| 2026-08-12 | 1 |  | 8 | yes | https://provider.engy.ai/api/subnet/v1/epoch/8 |
| 2026-08-19 | 5 |  | 9 | yes | https://provider.engy.ai/api/subnet/v1/epoch/9 |
| 2026-08-26 | 5 |  | 10 | yes | https://provider.engy.ai/api/subnet/v1/epoch/10 |
| 2026-09-02 | 7 |  | 11 | yes | https://provider.engy.ai/api/subnet/v1/epoch/11 |
| 2026-09-09 | 7 |  | 12 | yes | https://provider.engy.ai/api/subnet/v1/epoch/12 |
| 2026-09-16 | 8 |  | 13 | yes | https://provider.engy.ai/api/subnet/v1/epoch/13 |
| 2026-09-23 | 8 |  | 14 | yes | https://provider.engy.ai/api/subnet/v1/epoch/14 |

### Engy SN53: SN53 alpha bought back with revenue and burned, cumulative

Unit alpha · cumulative · source type chain · shape **rising** · key `engy-alpha-burned-cumulative`

Cumulative by construction. 18 burn rows on the team's buyback page, each linked to a taostats extrinsic (first 2026-08-12 19:24 UTC, 31.68 a; latest 2026-09-22 11:04 UTC, 1,416.88 a = $14,481.71). Rows sum to 10,823.62 a / $75,277; the page headline says 11,322.00 a total (one row did not parse, ~498 a). Third-party check: taorevenue.com shows SN53 30-day burn 271 t ($79,244) on 24 Sep 2026. Team says revenue funds the burns (Discord via SubnetRadar 13 and 17 Aug 2026).

| date | value | alpha | usd | verified | source |
|---|---:|---|---|---|---|
| 2026-08-12 | 31.68 | 31.68 | 199.77 | yes | https://taostats.io/extrinsic/8830493-0019?network=finney |
| 2026-08-12 | 1,171.50 | 1139.82 | 7154.56 | yes | https://taostats.io/extrinsic/8830570-0016?network=finney |
| 2026-08-13 | 1,231.98 | 60.48 | 363.14 | yes | https://taostats.io/extrinsic/8837947-0009?network=finney |
| 2026-08-14 | 1,319.09 | 87.11 | 514.3 | yes | https://taostats.io/extrinsic/8845446-0019?network=finney |
| 2026-08-16 | 1,393.35 | 74.26 | 466.95 | yes | https://taostats.io/extrinsic/8853819-0009?network=finney |
| 2026-08-17 | 1,461.43 | 68.08 | 443.2 | yes | https://taostats.io/extrinsic/8861740-0021?network=finney |
| 2026-08-20 | 1,607.40 | 145.97 | 950.4 | yes | https://taostats.io/extrinsic/8883132-0015?network=finney |
| 2026-08-23 | 1,755.77 | 148.37 | 1066.49 | yes | https://taostats.io/extrinsic/8910584-0025?network=finney |
| 2026-08-25 | 1,989.80 | 234.03 | 1527.2 | yes | https://taostats.io/extrinsic/8924743-0009?network=finney |
| 2026-08-27 | 2,393.21 | 403.41 | 2687.2 | yes | https://taostats.io/extrinsic/8932818-0012?network=finney |
| 2026-08-31 | 2,899.24 | 506.03 | 3539.22 | yes | https://taostats.io/extrinsic/8963188-0012?network=finney |
| 2026-09-02 | 3,825.68 | 926.44 | 5760.79 | yes | https://taostats.io/extrinsic/8976884-0013?network=finney |
| 2026-09-04 | 4,585.99 | 760.31 | 5175.0 | yes | https://taostats.io/extrinsic/8993688-0012?network=finney |
| 2026-09-07 | 5,431.21 | 845.22 | 5636.21 | yes | https://taostats.io/extrinsic/9012241-0002?network=finney |
| 2026-09-13 | 7,189.90 | 1758.69 | 10546.58 | yes | https://taostats.io/extrinsic/9057890-0003?network=finney |
| 2026-09-17 | 8,614.11 | 1424.21 | 9133.89 | yes | https://taostats.io/extrinsic/9085082-0008?network=finney |
| 2026-09-20 | 9,406.74 | 792.63 | 5630.71 | yes | https://taostats.io/extrinsic/9106103-0025?network=finney |
| 2026-09-22 | 10,823.62 | 1416.88 | 14481.71 | yes | https://taostats.io/extrinsic/9122715-0020?network=finney |

### Engy SN53: Dollar value of revenue-funded alpha burns per week

Unit USD/week · rate · source type chain · shape **rising** · key `engy-usd-burned-per-week`

Weekly sums (weeks ending Wednesday) of the burn table's USD column (the team's value at burn time). Week to 23 Sep: $29,246 across 3 burns, the high. Week to 12 Aug includes the first two burns ($7,354).

| date | value | alpha | burns | verified | source |
|---|---:|---|---|---|---|
| 2026-08-12 | 7,354.33 | 1171.5 | 2 | yes | https://provider.engy.ai/buyback (table with taostats extrinsic links) |
| 2026-08-19 | 1,787.59 | 289.93 | 4 | yes | https://provider.engy.ai/buyback (table with taostats extrinsic links) |
| 2026-08-26 | 3,544.09 | 528.37 | 3 | yes | https://provider.engy.ai/buyback (table with taostats extrinsic links) |
| 2026-09-02 | 11,987.21 | 1835.88 | 3 | yes | https://provider.engy.ai/buyback (table with taostats extrinsic links) |
| 2026-09-09 | 10,811.21 | 1605.53 | 2 | yes | https://provider.engy.ai/buyback (table with taostats extrinsic links) |
| 2026-09-16 | 10,546.58 | 1758.69 | 1 | yes | https://provider.engy.ai/buyback (table with taostats extrinsic links) |
| 2026-09-23 | 29,246.31 | 3633.72 | 3 | yes | https://provider.engy.ai/buyback (table with taostats extrinsic links) |

### SayGM SN28: Dollar traffic value served per day, monthly mean

Unit USD/day · rate · source type chain · shape **rising** · key `saygm-usd-traffic-per-day-monthly`

Built from the team's server-rendered per-epoch pages: 5 of every 20 daily epochs sampled (every 4th epoch), daily value = mean sampled epoch value x 20. Month only (Sep = 1-24 Sep; Jun = 26-30 Jun, first days with traffic). Gross traffic value at list price minus miner discount, i.e. GMV routed to Anthropic/OpenAI/Google/open models, not SayGM's take. Check: exact 20-epoch sum for 23-24 Sep on saygm.com/miners was $11,523 vs this estimate $11,844-12,468.

| date | value | sampled_days | verified | source |
|---|---:|---|---|---|
| 2026-06-01 | 359 | 5 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-01 | 2,111 | 31 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-01 | 7,836 | 31 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-24 | 12,976 | 24 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |

### SayGM SN28: Dollar traffic value served per day, weekly mean

Unit USD/day · rate · source type chain · shape **rising** · key `saygm-usd-traffic-per-day-weekly`

Same construction, weeks ending Wednesday. Rose ~5x from July to early September, then a plateau: 12,435 (9 Sep), 13,548 (16 Sep), 13,437 (23 Sep); the last point is 0.8% under the high, so strictly flat for three weeks. Daily high $16,081 on 10 Sep.

| date | value | verified | source |
|---|---:|---|---|
| 2026-07-01 | 311 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-08 | 1,702 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-15 | 1,592 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-22 | 2,510 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-29 | 2,526 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-05 | 4,016 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-12 | 7,451 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-19 | 7,627 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-26 | 10,016 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-02 | 9,886 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-09 | 12,435 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-16 | 13,548 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-23 | 13,437 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |

### SayGM SN28: Requests billed per day, weekly mean

Unit requests/day · rate · source type chain · shape **peaked** · key `saygm-requests-per-day-weekly`

Same construction. Peaked at 447K a day (week to 19 Aug), 251-290K in September; dollars kept rising because the mix moved to Claude Opus / Fable (higher value per request). 24 Sep partial day: 356K.

| date | value | verified | source |
|---|---:|---|---|
| 2026-07-01 | 44,878 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-08 | 129,283 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-15 | 140,962 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-22 | 194,402 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-29 | 181,709 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-05 | 225,856 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-12 | 321,064 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-19 | 447,380 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-26 | 430,365 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-02 | 418,987 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-09 | 274,508 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-16 | 251,188 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-23 | 290,508 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |

### SayGM SN28: Miners earning in an epoch (weekly max)

Unit miners · rate · source type chain · shape **rising** · key `saygm-miners-earning-per-epoch`

Header 'Miners' on the sampled epoch pages, max per week. 4 miners late June, 72 on 23 Sep 2026.

| date | value | verified | source |
|---|---:|---|---|
| 2026-07-01 | 7 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-08 | 5 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-15 | 4 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-22 | 9 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-07-29 | 9 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-05 | 18 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-12 | 29 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-19 | 49 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-08-26 | 59 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-02 | 47 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-09 | 49 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-16 | 47 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |
| 2026-09-23 | 72 | yes | https://saygm.com/miners/<epoch> (469 epoch pages sampled, every 4th epoch 23440-25314, header 'Traffic value / Requests / Miners') |

### DSperse (Inference Labs) SN2: zk proofs generated on SN2, cumulative

Unit proofs · cumulative · source type team · shape **rising** · key `dsperse-zk-proofs-cumulative`

Cumulative by construction. First three points are team posts; the last is read from the team's live API (sn2-api.inferencelabs.com/stats/: total_proofs 5,958,864,558; unique_miners 1,737; unique_validators 27; 24 Sep 2026). The 28 Jul figure is phrased 'Sertn has crossed 3.1B+ proofs' (Sertn is Inference Labs' platform; assumed the same counter). Implied rate: ~3.5M/day Nov 2025-Jun 2026, ~37M/day Jun-Jul 2026, ~49M/day Jul-Sep 2026; the team said peak 19.4M/day on 1 Jun. Caveats: the homepage showed a static '272 M total proofs' from Nov 2025 to Apr 2026 (Wayback) and shows '950 M' today, both inconsistent with the API; proofs are emission-funded miner work, not customer requests; SN2 7d emission ~0% and effective burn 94.66% over 20 epochs (team, 5 Sep 2026).

| date | value | verified | source |
|---|---:|---|---|
| 2025-11-13 | 300,000,000 | no | https://x.com/inference_labs/status/1989015447908626701 |
| 2026-06-01 | 1,000,000,000 | no | https://x.com/inference_labs/status/2061477893365284956 |
| 2026-07-28 | 3,100,000,000 | no | https://inferencelabs.com/media-room (entry 'Jul 28, 2026: Sertn has crossed 3.1B+ proofs') |
| 2026-09-24 | 5,958,864,558 | yes | https://sn2-api.inferencelabs.com/stats/ |

## Single numbers

| subnet | netuid | metric | value | unit | date | source | type | note |
|---|---|---|---:|---|---|---|---|---|
| Chutes | 64 | GPUs online | 4,312 | GPUs | 2026-09-24 | https://chutes.ai/api/research/overview | chain | Also gpu_types 22, active_instances 217, requests_last_hour 125977. Same 4,312 on 17 Sep (roster) so no movement in a week; api.chutes.ai/nodes/ lists 3,661 idle + 651 provisioned GPU slots across 22 types. No public history; team said on 23 Sep the token drop was due to 'fewer GPUs and slower models' (Discord via SubnetRadar). |
| Chutes | 64 | Public chutes (deployed models/apps) | 491 | chutes | 2026-09-24 | https://api.chutes.ai/chutes/?limit=100&include_public=true (total 491) | chain | All 491 are TEE; 476 vllm template; 31 have live instances; created_at is reset on redeploy so it cannot give a history (rows date only from Oct 2025). arXiv trace: 9,174 models over the year (3,922 public, 5,252 private). |
| Chutes | 64 | Revenue per trillion LLM tokens | 340,000 | USD per trillion tokens | 2026-07-20 | https://x.com/chutes_ai/status/2079637387622383955 (Week 10; article x.com/i/article via status 2079637219049189642) | team | All-time high as of 2026-07-20; 90-day average $282K, 'climbing since we started tracking it'. Ratio, not a dollar total. No later value posted (Weeks 11-19 articles carry no revenue figures). |
| Chutes | 64 | Revenue, trailing 90 days | 1,130,000 | USD | 2026-08-30 | https://subnetradar.com/subnet-news/64/2026-08-30 (team in Discord) | team | $4.58M annualised; of it $214.6K subscriptions; 24,408 new subscribers in the 90 days. Team dashboard figure relayed by SubnetRadar; the 'daily revenue page' itself was not found (it was offline 20-21 Aug per Discord). |
| Chutes | 64 | Revenue-funded alpha burned, cumulative | 260,770 | alpha | 2026-08-31 | SubConnect via docs/research/top-subnets-2026.md (burn_alpha extrinsics) | third-party | Month only. Also 20,944.85 a mid-Jul to mid-Aug 2026. Team confirmed 14 Sep the buyback wallet 'has not been used in weeks' (Discord via SubnetRadar); taorevenue.com 30-day burn for SN64 = 35 t ($10,114) on 24 Sep 2026, versus 4,544 t for Lium. Auto-buyback reported down on 18 Aug. |
| Chutes | 64 | Users in the one-year production trace | 314,970 | users | 2026-04-12 | https://arxiv.org/abs/2608.13573 Table 2 | team | Trace 2025-04-11 to 2026-04-12: 6,122,413,756 requests, 35.80T input tokens, 2.52T output tokens, 9,174 models, 875,921 serving instances. The paper gives no monthly table; Fig. 1 charts daily requests (rises to a late-summer/fall 2025 peak, declines into 2026) and active models per day (<100 to >400). |
| Chutes | 64 | Free-tier cut, effect stated by team | -45 | percent | 2026-03-01 | docs/research/top-subnets-2026.md (team statement, month only) | team | Tokens -45%, revenue -25% after cutting free-tier traffic in March 2026; revenue covers 'thirty to forty percent of operating expenses' (SubConnect quoting team, Aug 2026). No paid/free share series published. |
| Engy | 53 | Total alpha burned (page headline) | 11,322.0 | alpha | 2026-09-24 | https://provider.engy.ai/buyback | chain | Table rows sum to 10,823.62 a; see series engy-alpha-burned-cumulative. |
| Engy | 53 | Epoch 15 in progress (23-30 Sep) | 4,837.33 | USD settled | 2026-09-24 | https://provider.engy.ai/api/subnet/v1/epoch/15 | chain | After ~1.4 days: 1,494,892 requests, 15.6B tokens, 80 miners, 8 models. Not final. |
| Engy | 53 | Kimi K3 inference price increase | 30 | percent | 2026-08-29 | https://subnetradar.com/subnet-news/53/2026-08-29 | third-party | Pricing page today: deepseek-v4.1-flash $0.04/$0.08 per M in/out; deepseek-v4-flash-0731 $0.045/$0.09; qwen3.6-35b-a3b $0.045/$0.30; kimi-k3 rate card 1,950 micro/1k input in the epoch params. No dated price history found. |
| Engy | 53 | Named customer | 1 | customers | 2026-08-04 | https://x.com/totheagi/status/2084650424792121659 | team | Almanac (a Bittensor subnet team) welcomed as a user. No revenue figures posted by @engyai or @totheagi (3 tweets matched revenue/customer terms since Jul 2026). |
| SayGM | 28 | Network tokens processed, cumulative | 96,480,000,000 | tokens | 2026-09-07 | https://x.com/say_gm_/status/2096886191081615619 | team | >30B of it output tokens. Not on the pages; team post only. |
| SayGM | 28 | August token volume vs July | 545 | percent | 2026-08-01 | https://x.com/say_gm_/status/2096886191794668016 | team | Month only. Output tokens +412% to 20.44B in August 2026. |
| SayGM | 28 | Models listed | 56 | models | 2026-09-24 | https://saygm.com/models (56 /models/<slug> links) | chain | Roster had 47 with data on 17 Sep; no history page. |
| SayGM | 28 | Alpha burned, 30 days | 519 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $151,626. taorevenue's 'burn' column counts on-chain alpha burns; SayGM has not published a burn ledger. Team said 30 Jul: 'hit 50% miner burn through actual usage'. |
| SayGM | 28 | Latest finalised epoch | 722.58 | USD traffic value | 2026-09-24 | https://saygm.com/miners (epoch 25,314, 16:36 UTC) | chain | 27,374 requests, 72 miners, 2 gateways in the 72-minute epoch. |
| Verathos | 96 | Requests served, cumulative | 1,485,085 | requests | 2026-09-24 | https://verathos.ai/api/dashboard (usage_stats) | chain | Input tokens 2,632,923,762, output 344,381,737; last 24h: 4,916 requests, 1,107,262 tokens, active_users_24h 1. Paid: x402_requests 1735, x402_usdc_total $0.12, total_consumed_usd $0.03. Mostly validator/synthetic traffic on Qwen3.5-9B; essentially no paying use. 80 healthy miners, 12 models, 5 validators; 294 miner rows in the dashboard. |
| Verathos | 96 | Team token sales for expenses | 10,000 | USD/month | 2026-09-16 | https://subnetradar.com/subnet-news/96/2026-09-16 | third-party | Team member disclosure in Discord; Sleipnir mesh inference live on mainnet since 4 Sep 2026. |
| DSperse (Inference Labs) | 2 | Effective emission burn rate, 20-epoch average | 94.66 | percent | 2026-09-05 | https://subnetradar.com/subnet-news/2/2026-09-05 | third-party | Miners received 5.34% of nominal emissions on average; team says intentional. |
| DSperse (Inference Labs) | 2 | Unique proving miners | 1,737 | miners | 2026-09-24 | https://sn2-api.inferencelabs.com/stats/ | chain | Homepage says 1,402 proving nodes; 1B-proofs graphic (1 Jun) said 1.7K active miners, 24 validators. |
| SOMA | 114 | CoT-Compression-10 competition | 20 | uploads | 2026-09-24 | https://thesoma.ai/dashboard | chain | Upload window 22-29 Sep; prize pool 715.48 a (41,328 in alpha units shown); 15 qualified. Dendrite gateway has no public token counter (thesoma.ai/dendrite 404). |
| Cacheon | 14 | Kernel arena Qwen baseline | 1,400 | tokens/s | 2026-09-24 | https://subnetradar.com/subnet-news/14/2026-09-24 | third-party | Cacheon is now a GPU-kernel competition (beat sglang, earn up to 33 TAO/day), not an inference cache; no usage counter on cacheon.ai; /dashboard 404. |
| Instant | 46 | Named customer agreements | 1 | customers | 2026-08-14 | https://subnetradar.com/subnet-news/46/2026-08-14 | third-party | Ditto first; launch still pending as of 22 Sep 2026 (waitlist page only at instantsubnet.com). No usage numbers. |
| DSperse | 2 | Alpha burned, 30 days (taorevenue.com) | 244 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $71,276. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |
| Instant | 46 | Alpha burned, 30 days (taorevenue.com) | 323 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $94,242. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |
| Actual | 95 | Alpha burned, 30 days (taorevenue.com) | 3,768 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $1,099,840. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |
| Verathos | 96 | Alpha burned, 30 days (taorevenue.com) | 223 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $65,064. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |
| Cacheon | 14 | Alpha burned, 30 days (taorevenue.com) | 74 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $21,654. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |
| MicroTensor | 92 | Alpha burned, 30 days (taorevenue.com) | 138 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $40,365. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |
| Chutes | 64 | Alpha burned, 30 days (taorevenue.com) | 35 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $10,114. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |
| Engy | 53 | Alpha burned, 30 days (taorevenue.com) | 271 | TAO equivalent | 2026-09-24 | https://taorevenue.com | third-party | $79,244. Includes registration/recycle and mechanism burns as well as revenue buybacks; Actual's 3,768 t with 0% emission and no product is not customer revenue. SN104 and SN114 show no burn. |

## Not rising

- `chutes-new-users-per-month` (falling): New user registrations per month, Chutes SN64.
- `chutes-image-requests-per-day` (peaked): Image-generation (diffusion) requests per day, monthly mean, Chutes SN64.
- `engy-miners-paid-per-epoch` (peaked): Miners in the settlement ledger per epoch, Engy SN53.
- `saygm-requests-per-day-weekly` (peaked): Requests billed per day, weekly mean, SayGM SN28.
- Also: Chutes tokens/requests a day (done elsewhere) are down ~80% from the 2025 peak; Chutes burns have stopped (35 t in 30 days); Verathos has ~$0.15 of lifetime paid usage; SN2 proofs rise but emission to miners is ~0%.

## Could not find

- Chutes GPUs-online history: only the live overview tile (4,312 today, 4,312 on 17 Sep); research bundle exposes only type=llm and type=users; Wayback has the client-rendered page without numbers; no team chart.
- Chutes public-chutes/models count over time: chute created_at resets on redeploy; arXiv Fig. 1c is the only history (image saved).
- Chutes revenue over time: Weeks 9-19 'Last Week in Chutes' articles (fxtwitter) carry no dollar figures except the 20 Jul $340K/trillion ratio; the team's 'daily revenue page' referenced in Discord was not located; 90-day $1.13M is one point.
- Chutes revenue-funded burn_alpha extrinsic list: taomarketcap has no burns page; not read from chain (would need extrinsic scans); SubConnect totals used as singles.
- Chutes paid vs free share by day: not published; team statements only.
- arXiv 2608.13573 monthly users/requests tables: paper has figures, not tables (Table 2 is a one-year total).
- Engy customer/revenue posts: 3 tweets matched since Jul 2026, none with figures; engy.ai pricing page has no change log.
- SayGM credits sold / SayGM's own revenue: not published; only gross traffic value per epoch.
- SayGM models-listed history: no archive; 56 today vs 47 with data on 17 Sep (roster).
- SayGM monthly volume posts before September: X budget stopped at 23 reads; the two 7 Sep posts were read via fxtwitter.
- Verathos proofs count and alpha-locked total: /api/dashboard has requests/tokens only; conviction endpoints need a session; no history.
- Cacheon SN14, Instant SN46, Actual SN95, TAOstatus SN104, MicroTensor SN92: no usage counters, dashboards or dated numbers on their sites (taostatus.ai and microtensor.io do not resolve; taostatus.com and microtensor.xyz are landing pages).
- SOMA SN114 Dendrite tokens routed: no counter; only the compression-competition dashboard.

## X reads used

23 of 30 (log: `/tmp/sweep/xreads-inference.log`; 3 calls: two `from:chutes_ai` searches returning 10 each, one `from:engyai OR from:totheagi` returning 3). Everything else came from fxtwitter (free), team APIs and pages, SubnetRadar briefs, taorevenue, Wayback.

## Charts saved

| file | source URL |
|---|---|
| `media/charts-gallery/chutes-requests-per-day-arxiv.png` | https://arxiv.org/html/2608.13573v2/fig_request_rate_day.png (Fig. 1a, daily requests Apr 2025-Apr 2026) |
| `media/charts-gallery/chutes-active-models-per-day-arxiv.png` | https://arxiv.org/html/2608.13573v2/fig_daily_unique_models.png (Fig. 1c, active models per day) |

Not saved (not charts): SayGM '100 Billion Tokens' promo graphic (status 2096886191081615619), Inference Labs '300 Million' and '1B' milestone graphics; Engy and SayGM publish tables, not chart images.

## Strongest rising series, for the coordinator

1. `engy-usd-settled-per-epoch`: $682 (week to 29 Jul) to $34,870 (week to 23 Sep), signed ledger, 9 weekly points; miners fell 199 to 82 while dollars rose 8x.
2. `saygm-usd-traffic-per-day-monthly`: ~$360/day (late Jun) to $2.1K (Jul), $7.8K (Aug), $13.0K (Sep to the 24th), read from 469 epoch pages; gross routed value, plateau since 9 Sep.
3. `engy-alpha-burned-cumulative` / `engy-usd-burned-per-week`: 0 to 10,824 a ($75K) in six weeks with extrinsic links; $29K burned in the week to 23 Sep.
4. `dsperse-zk-proofs-cumulative`: 300M (Nov 2025) to 5.96B (24 Sep 2026, live API), but emission-funded work with ~0% paid to miners.
5. `chutes-users-cumulative`: 802K registrations, rising by construction; the rate (`chutes-new-users-per-month`) is falling.