Published July 9, 2026 · data June 20, 2026 → August 27, 2026 (69 daily snapshots) · live edition — extends with each daily snapshot until the next report ships
What running AI actually cost this month — computed straight from Spanvero's append-only daily price record, the same data you can download. No estimates, no smoothing: every figure carries its date.
Daily snapshots 69 Jun 20 → Aug 27 | Prices logged 36,164 rows appended to the log | Models tracked 412 live market, latest day | Real price moves 60 34 cuts · 26 rises |
Between June 20, 2026 and August 27, 2026, Spanvero logged 36,164 prices across 69 daily snapshots: the full live market listing (412 priced models on the latest day), 85 GPU rental rates, and the dated per-model prices Spanvero publishes. The log is append-only — a day, once written, is never revised.
The floor moved. The cheapest paid model went from $0.02 per 1M blended tokens on June 20, 2026 (inclusionAI: Ling-2.6-flash) to $0.0245/1M on August 27, 2026 (Mistral: Mistral Nemo) — up 23%.
Above the floor, prices did move: 60 real day-over-day changes cleared our 0.5% noise threshold — 34 cuts against 26 rises. The largest single move was Meta: Llama 3.3 70B Instruct, rising from $0.21 to $0.71/1M on August 26, 2026 (+238%); the largest cut was DeepSeek: DeepSeek V4 Pro 0423, from $2.4 to $0.646/1M on August 22, 2026 (−73%).
On the rental side, the cheapest tracked GPU went from $0.26/hr on June 20, 2026 (NVIDIA RTX 3090 24GB) to $0.06/hr on August 27, 2026 (NVIDIA RTX 3060 12GB). Read that with care: our GPU coverage grew from 10 to 85 SKUs over the window, and the new floor comes from a card that entered the record mid-window — so much of that drop is wider tracking, not vendor repricing. Like-for-like — only the SKUs tracked for the whole window — the floor held at $0.26/hr (NVIDIA RTX 3090 24GB).
Finally, the market ended the window 76 models larger than it began (entries minus exits), and truly-free endpoints ($0 in, $0 out) went from 27 to 21.
Single-day changes on the live market listing, largest first — 60 in this window. The full feed lives on the price record.
| Name | Before | After | Change | Date |
|---|---|---|---|---|
| Meta: Llama 3.3 70B Instruct | $0.21/1M | $0.71/1M | +238% | Aug 26 |
| Qwen: Qwen3 235B A22B Instruct 2507 | $0.32/1M | $0.875/1M | +173% | Aug 27 |
| DeepSeek: DeepSeek V4 Flash Vision Exp | $0.44/1M | $0.88/1M | +100% | Aug 25 |
| DeepSeek: DeepSeek V3.1 | $0.6/1M | $1.1/1M | +83% | Aug 22 |
| MoonshotAI: Kimi K2.6 | $1.4108/1M | $2.475/1M | +75% | Aug 23 |
| DeepSeek: DeepSeek V4 Pro 0423 | $2.4/1M | $0.646/1M | −73% | Aug 22 |
| DeepSeek: DeepSeek V4 Flash 0731 | $0.13/1M | $0.21/1M | +62% | Aug 24 |
| DeepSeek: DeepSeek V4 Flash 0731 | $0.21/1M | $0.09/1M | −57% | Aug 26 |
| DeepSeek: DeepSeek V4 Flash 0423 | $0.0861/1M | $0.1329/1M | +54% | Aug 25 |
| DeepSeek: DeepSeek V4 Pro 0423 | $0.7891/1M | $1.1856/1M | +50% | Aug 25 |
Window-start vs window-end change in the dated price Spanvero publishes per model — each links to that model's live cost page. 31 of 70 tracked models moved.
| Name | Before | After | Change | Last changed |
|---|---|---|---|---|
| Qwen3 30B A3B Thinking 2507 (entered Jun 23) | $0.24/1M | $1.3/1M | +442% | Aug 4 |
| Hy3 preview (entered Jun 23) | $0.1365/1M | $0.39/1M | +186% | Aug 18 |
| Llama 3.1 8B Instruct | $0.025/1M | $0.065/1M | +160% | Jul 21 |
| Gemma 3 27B (entered Jun 23) | $0.12/1M | $0.265/1M | +121% | Jul 28 |
| Qwen2.5 7B Instruct | $0.07/1M | $0.15/1M | +114% | Aug 4 |
| DeepSeek V4 Flash 0731 (entered Aug 4) | $0.135/1M | $0.21/1M | +56% | Aug 18 |
| Qwen3 Next 80B A3B Thinking (entered Jul 7) | $0.4388/1M | $0.675/1M | +54% | Aug 4 |
| DeepSeek V4 Flash (entered Jun 23) | $0.135/1M | $0.0861/1M | −36% | Aug 25 |
| Hy3 (entered Jul 21) | $0.5/1M | $0.33/1M | −34% | Jul 28 |
| DeepSeek-V3 | $0.5/1M | $0.6432/1M | +29% | Aug 4 |
Hourly rates that changed over the window, per tracked SKU — each links to that card's rate page.
| Name | Before | After | Change | Last changed |
|---|---|---|---|---|
| NVIDIA A100 40GB | $1.10/hr | $1.99/hr | +81% | Jul 3 |
| NVIDIA H200 141GB | $3.49/hr | $4.39/hr | +26% | Jul 3 |
| NVIDIA RTX A5000 24GB | $0.36/hr | $0.27/hr | −25% | Jul 3 |
| NVIDIA H100 80GB SXM | $2.69/hr | $3.29/hr | +22% | Jul 3 |
| NVIDIA L40S 48GB | $0.86/hr | $0.99/hr | +15% | Jul 3 |
| NVIDIA L4 24GB | $0.43/hr | $0.39/hr | −9.3% | Jul 3 |
Sources, labeled: the market floor and single-day moves come from the live OpenRouter listing (the full market); published-model rows are the dated prices Spanvero publishes; GPU rates are the vendors' own posted prices. We sell no compute and mark nothing up. How we stay honest →
Free to quote with a link. The underlying record: history.json · history.csv · prices.json · gpus.json · the full feed →
Spanvero, “State of Inference Costs — July 2026” (data 2026-06-20 → 2026-08-27) — spanvero.com/reports/inference-costs-2026-07/
Want the Outcome Lab run against your own evidence? Flat-fee Outcome Cost audits are $49 / $249 / $499. How the audits work →
All reports → · The daily price record · Outcome Lab →
Data June 20, 2026 → August 27, 2026; published July 9, 2026. $0 markup, your own accounts, we never resell compute. © 2026 Cynosure LLC.
A short email of real AI price moves, straight from the daily log — no hype. We're collecting the list now; the first issue goes out when it opens. Unsubscribe with one click.