Entry · RTX 2060
High supplyReference anchor1650 · 822 offers
On platform:823 units
Public data · updated every 30 minutes ⓘ
Global market price band vs. Zhimao internal price (last 90 days, incl. real trade snapshots).
Anonymized aggregate of listed prices across major global compute markets (China region and overseas region included), in USD per GPU-hour, refreshed every 6 hours.
| Tier | GPU model | China region ($/hr) | Overseas region ($/hr) | Zhimao internal | 7-day trend |
|---|---|---|---|---|---|
| compute.market.tierShort.h20 | H20 | — | $0.900 – $1.40 | $0.528 | |
| compute.market.tierShort.b300 | B300 | — | $6.94 – $12.50 | $2.80 | |
| Entry | Tesla T4 | — | $0.120 – $0.174 | $0.046 | |
| Mainstream | RTX 3060 | $0.081 | $0.058 – $0.900 | $0.094 | |
| Flagship | RTX 3090 | $0.189 – $0.194 | $0.148 – $0.240 | $0.161 | |
| Enthusiast | RTX 4090 | $0.308 – $0.321 | $0.280 – $0.490 | $0.230 | |
| Pro | RTX A6000 | $0.810 – $0.881 | $0.330 – $1.97 | $0.585 | |
| Datacenter | H100 SXM | — | $1.45 – $3.51 | $0.975 | |
| Ultra | H200 | — | $0.500 – $4.67 | $1.89 |
Some markets expose no public API; curated reference prices (periodically verified) are included for comparison only.
🌐 External rental
Plug your GPU into the Global Compute Net for USD; earn top prices when demand is high.
🔄 Zhimao platform tasks
Take Zhimao AI tasks (drama / short video / lead-gen) during idle hours: stable, guaranteed income, never idle for free; Credits are withdrawable/redeemable.
Tell me your GPU and uptime, and I'll estimate your monthly Credits.
External band: live global price intel (last 48h).
Higher utilization = more external rental income; idle hours auto-run internal fallback tasks.
Estimated Monthly Earnings
1,307 ~ 1,758Credits/mo
≈ ¥187 ~ ¥251 / mo
💡 Idle hours auto-run internal platform tasks (inference / harvest / media) — a built-in earnings floor. Other compute platforms require you to build idle jobs yourself.
Basis: external rental ≈ market median for the same model × ~51% net host share; internal ≈ 0.000025 Credits/token × typical throughput, conservatively assuming idle hours ~50% filled. Payout ref: 50 Credits ≈ $1.
Estimates only, not a guarantee. Real earnings = price × utilization, depending on market demand, uptime and device reputation.
⊙ Supply–demand index
Balanced — stable earnings
On-demand listed prices aggregated across 70+ global cloud GPU providers, synced every 6 hours. Updated 8月29日 00:01
| GPU model | VRAM | Global avg | Price range | Providers |
|---|---|---|---|---|
| H100 SXM | 80G | $3.51 | $1.68 – $11.06 | 44 |
| H200 | 141G | $4.67 | $1.40 – $10.60 | 34 |
| A100 SXM | 80G | $1.97 | $0.580 – $4.10 | 33 |
| L40S | 48G | $1.49 | $0.380 – $3.77 | 28 |
| RTX PRO 6000 | 96G | $2.29 | $0.190 – $7.94 | 27 |
| B200 | 192G | $6.46 | $3.69 – $14.00 | 21 |
| RTX A6000 | 48G | $0.820 | $0.280 – $2.03 | 19 |
| HGX B300 | 288G | $10.05 | $5.90 – $37.51 | 17 |
| L4 | 24G | $0.900 | $0.170 – $1.67 | 16 |
| A100 PCIE | 40G | $1.50 | $0.270 – $3.00 | 15 |
| RTX 6000 Ada | 48G | $1.02 | $0.600 – $1.57 | 15 |
| L40 | 40G | $1.02 | $0.330 – $2.50 | 15 |
| Tesla V100 | 32G | $1.46 | $0.110 – $2.57 | 12 |
| RTX 4090 | 24G | $0.470 | $0.160 – $0.740 | 12 |
| RTX 5090 | 32G | $0.670 | $0.250 – $0.990 | 11 |
Averages are the median of cross-provider on-demand listed prices; spot/reserved run lower. For reference only — actual rates vary by provider.
Market reference price and two earning lenses per GPU tier, distilled from live global compute market data (updated 2026-07-03). Live rates actually move very little, so these are stable median bands, not extreme highs/lows.
Entry · RTX 2060
High supplyReference anchor1650 · 822 offers
On platform:823 units
Mainstream · RTX 3060
High supplyReference anchor3060 · 1,253 offers
On platform:1,255 units
Flagship · RTX 4070
High supplyReference anchor3080 · 939 offers
On platform:939 units
Datacenter · H100
High supplyReference anchorH100 · 157 offers
On platform:157 units
🔄 Zhimao platform tasks · stable income
Platform task income is stable and low-volatility, filling idle hours without competing with your high-price external rentals.
Put your GPUs on the network?
Pick a path: onboard your own hardware, or buy a Zhimao Box. Listing must be enabled before your machine appears on the rental shelf.
Supply/utilization from real telemetry; prices are market anchors. Real earnings = price × occupancy, subject to supply/demand and device reputation.