Last Updated
2026-06-30
Compute Price Tracker
The fastest way to read compute price trends
This page turns the project's database, daily report, and charts into one lightweight entry point. It is designed for quick reading of price trends, latest snapshots, and single-model history.
Days Tracked
31
Models Covered
162
Chart Index
63 + 12
Database Records
66105
Lead Signals
Read each block in order: Vast.ai GPU supply and price movement first, then TPU / ASIC Spot signals. Percentage-change charts come before absolute price charts where available.
Market Tightness Readout
Marketplace signal is mixed
Vast.ai high-end cluster price and supply are not moving in the same direction yet. Add AWS Capacity Blocks credentials for a stronger reservable-cluster signal.
Vast 8+ GPU Supply
280 GPUs
Vast Median Price
$2.1336
Vast Supply Day Change
-27.1%
Vast Price Day Change
+45.4%
Reliable Supply Share
74.3%
AWS Availability Rate
N/A
AWS 64+ Unavailable
N/A
AWS Earliest Wait
N/A
AWS Query Count
N/A
Core Compute Supply Pressure
A fixed representative supply-pressure signal for each core accelerator. This is the top-level market temperature read.
Core supply-pressure relative change
Percentage change versus the first observation. This is the fastest read for which compute classes are loosening or tightening.
Core supply-pressure absolute price
Absolute price for the same fixed representative signal per accelerator.
Professional AI GPUs
Data-center and workstation-style GPUs that better reflect core AI infrastructure pricing.
Vast.ai professional AI GPU supply trend
Supply comes first across the professional AI GPUs leading the Vast.ai marketplace.
Vast.ai professional AI GPU marketplace median % change
Then relative price change versus the first observation date for the same professional AI GPUs.
Vast.ai professional AI GPU marketplace median trend
Then the absolute marketplace median price for the same professional AI GPUs.
AI Market-Active GPUs
Consumer-heavy GeForce GPUs that remain highly active in the AI rental market.
Vast.ai AI market-active GPU supply trend
Supply comes first across the GeForce-style GPUs that are most active in the AI market.
Vast.ai AI market-active GPU marketplace median % change
Then relative price change versus the first observation date for the same market-active GPUs.
Vast.ai AI market-active GPU marketplace median trend
Then the absolute marketplace median price for the same market-active GPUs.
High-End Cluster Supply
This block tracks whether complete high-end clusters are easing or still scarce. Vast.ai shows 8+ GPU marketplace supply; AWS Capacity Blocks show future reservable cluster availability when AWS credentials are configured.
Vast.ai high-end 8+ GPU cluster supply trend
Tracks rentable GPUs inside high-end 8+ GPU Vast.ai offers, not all single-card supply.
Vast.ai high-end 8+ GPU cluster median % change
Relative cluster price movement versus the first collected 8+ GPU cluster observation.
Vast.ai high-end 8+ GPU cluster median trend
Absolute median price for high-end 8+ GPU Vast.ai cluster offers.
Vast.ai reliable high-end cluster supply share
Share of high-end 8+ GPU supply with reliability at or above the configured threshold.
AMD Market Proxy
Competitive AMD listed prices and Azure Spot summaries. This is a market proxy, not a true exchange order book.
AMD market proxy % change
Relative movement versus the first tracked observation, shown before absolute price for faster signal reading.
AMD market proxy absolute price
Public competitive AMD GPU pricing from listed neo-cloud prices plus Azure Spot summary lines.
AMD GPU Spot Baseline
Azure AMD MI300X Spot pricing is a public spot-rate signal, not a marketplace supply feed.
Azure AMD MI300X Spot regional range
Shows the lowest, median, and highest Azure region price so one stable low region does not hide regional spread.
Azure AMD MI300X Spot key regions
Tracks representative low, mid, and high-price Azure regions for AMD MI300X Spot pricing.
Google Cloud Hyperscaler Signals
Google Cloud core GPU / TPU prices are closer to large-model cloud rental economics than long-tail marketplaces.
Google Cloud GPU Spot movers relative change
Only GPU Spot models whose observed price actually moved, ordered by movement magnitude.
Google Cloud GPU Spot movers absolute price
Absolute Spot prices for the same moving GPU models, separated from high-price DWS lines.
Google Cloud On-demand price trend
Core Google Cloud GPU / TPU listed prices. This is the stable official cost anchor.
Google Cloud On-demand relative change
Relative movement versus the first tracked day for Google Cloud core accelerators.
Google Cloud Spot / DWS supply signal
Spot and DWS prices are shown as separate lines because they are different public supply-pressure signals.
Google Cloud Spot / DWS relative change
Percentage change for Spot and DWS lines without mixing price types.
Google Cloud CUD commitment price
3-year CUD price trend for long-term customer cost anchoring.
Google Cloud CUD relative change
3-year CUD percentage movement versus first tracked day.
TPU / ASIC Signals
Google TPU Spot moves after the GPU market blocks, with relative movement before absolute price.
Google TPU Spot relative change
Relative change versus the first collected TPU Spot observation. This is the fastest read for TPU price movement.
Google TPU Spot price trend
Daily lowest-region TPU Spot price by model after the relative-change view.
Google TPU Spot snapshot
Latest lowest-region TPU Spot price by model, with the number of regions observed.
| Focus Model | Latest Price | Vs First Day | Vs Previous Day | Baseline |
|---|---|---|---|---|
| NVIDIA GB300 | $18.0000 | 0.0% | 0.0% | On-demand |
| NVIDIA GB200 | $16.0000 | 0.0% | 0.0% | On-demand |
| NVIDIA B300 | $15.0000 | 0.0% | 0.0% | On-demand |
| NVIDIA B200 | $8.0550 | -42.5% | 0.0% | On-demand |
| NVIDIA H200 | $4.5753 | -42.2% | 0.0% | On-demand |
| NVIDIA H100 | $4.2008 | -38.9% | 0.0% | On-demand |
| AMD MI355X | $8.6000 | 0.0% | 0.0% | On-demand |
| AMD MI300X | $6.0000 | 0.0% | 0.0% | On-demand |
| AWS Trainium2 | Capacity Block | $2.2350 | 0.0% | 0.0% | Capacity Block |
| AWS Trainium | $1.3438 | 0.0% | 0.0% | On-demand |
| Google Ironwood | $12.0000 | 0.0% | 0.0% | On-demand |
| Google Trillium | $2.7000 | 0.0% | 0.0% | On-demand |
| Google TPU v5p | $4.2000 | 0.0% | 0.0% | On-demand |
Latest Moves
This section highlights changes between the two latest collection dates and the most supplied Vast.ai models.
Rising Recently
| Item | Latest Price | Change |
|---|---|---|
| No rising items were detected between the two latest collection dates. | ||
Pulling Back
| Item | Latest Price | Change |
|---|---|---|
| No pullbacks were detected between the two latest collection dates. | ||
Top Vast Supply
| Model | Supply | Median Price |
|---|---|---|
| RTX 4090 | 151 GPUs | $0.9007 |
| RTX 5080 | 99 GPUs | $0.2815 |
| H100 SXM | 94 GPUs | $5.8678 |
| RTX PRO 6000 S | 84 GPUs | $5.3363 |
| H200 | 83 GPUs | $7.6371 |
| B200 | 80 GPUs | $17.3944 |
| RTX 3080 Ti | 62 GPUs | $0.1659 |
| RTX 5060 Ti | 61 GPUs | $0.2689 |
Model Explorer
This is the fastest path to a single model. Search or filter on the left, then read the chart on the right.
No models match the current filters.
NVIDIA GB300
NVIDIA | GPU
Cross-provider price curve
On-demand
Latest Price
$18.0000
Vs Previous Day
0.0%
Vs First Day
0.0%
Latest Date
2026-06-30
Providers: oracle_cloud
Latest price uses the lowest observed value for this model across providers and price buckets on the latest date.
Snapshot Overview
If you only want a quick scan of the latest prices and supply, this section is the fastest read.
Core accelerator price trend
Provider-specific official baseline for each core accelerator. Lines do not switch between cloud providers.
Core accelerator relative change
Change versus the first observation date using the same provider per accelerator.
Official price snapshot
Compare the latest lowest official prices across major cloud providers.
AI GPU market snapshot
Current price distribution across CoreWeave, RunPod, Lambda, and Vast.ai.
Vast.ai supply snapshot
Estimated rentable GPU supply and current market prices for top models.
Official Lows
| Provider | Model | Price |
|---|---|---|
| google_cloud | NVIDIA T4 | $0.3500 |
| google_cloud | NVIDIA L4 | $0.5582 |
| aws | AWS Trainium | $0.5960 |
| google_cloud | NVIDIA P4 | $0.6000 |
| aws | AWS Trainium | $1.3438 |
| google_cloud | NVIDIA P100 | $1.4600 |
| aws | NVIDIA A100 40GB | $1.4750 |
| google_cloud | NVIDIA A100 | $1.6136 |
| aws | NVIDIA L4 | $1.6688 |
| aws | NVIDIA A100 80GB | $1.8450 |
| google_cloud | NVIDIA A100 80GB | $1.8462 |
| aws | NVIDIA A10G | $2.0360 |
Market Lows
| Provider | Model | Price |
|---|---|---|
| coreweave | NVIDIA L40 | $0.1562 |
| runpod | NVIDIA RTX A5000 | $0.1600 |
| vast.ai | RTX 3080 Ti | $0.1659 |
| runpod | NVIDIA RTX 3090 | $0.2200 |
| vast.ai | RTX 5060 Ti | $0.2689 |
| runpod | NVIDIA RTX A5000 | $0.2700 |
| coreweave | NVIDIA L40S | $0.2812 |
| vast.ai | RTX 5080 | $0.2815 |
| vast.ai | RTX 5070 Ti | $0.2956 |
| coreweave | NVIDIA RTX PRO 6000 Blackwell Server Edition | $0.3125 |
| vast.ai | RTX 3090 | $0.3222 |
| runpod | NVIDIA RTX A6000 | $0.3300 |
Vast Supply
| Model | Supply | Median Price |
|---|---|---|
| RTX 4090 | 151 GPUs | $0.9007 |
| RTX 5080 | 99 GPUs | $0.2815 |
| H100 SXM | 94 GPUs | $5.8678 |
| RTX PRO 6000 S | 84 GPUs | $5.3363 |
| H200 | 83 GPUs | $7.6371 |
| B200 | 80 GPUs | $17.3944 |
| RTX 3080 Ti | 62 GPUs | $0.1659 |
| RTX 5060 Ti | 61 GPUs | $0.2689 |
| A100 SXM4 | 61 GPUs | $2.6433 |
| RTX PRO 6000 WS | 60 GPUs | $2.3520 |
| RTX 5070 Ti | 58 GPUs | $0.2956 |
| L40S | 52 GPUs | $1.1341 |
Long-Term Trends
Daily data is useful for direction. Quarterly and yearly charts are better for price baselines and supply inflection points.
Official GPU quarterly trend
Compress daily data into quarterly points for price-baseline tracking.
Official GPU yearly trend
A longer view of how price levels are moving over time.
Official TPU quarterly trend
Longer-horizon movement across Google TPU generations.
Vast.ai quarterly price trend
Quarterly trend for public GPU marketplace prices.
Vast.ai quarterly supply trend
Quarterly change in estimated rentable public GPU supply.
