Loading prices…
Trainium2 rental prices
AWS Trainium · Trainium · 96 GB HBM3 · Trainium · 2024
1 providers · 1 configurations · prices as of Nov 9, 2026, 21:02 UTC
Lowest on-demand
—per GPU·h
Median on-demand
—
across 0 providers
Lowest spot
$0.54/GPU·h
interruptible
FP16/BF16 cost
—
Providers
1
1 configurations
Price history
Daily Trainium2 prices per GPU-hour. The shaded period is reconstructed from public archives: the version history of SkyPilot's price catalogs, Internet Archive copies of providers' pricing pages and the open gpu-price-feed project (fewer providers than live data). After that, PetaPrice's own tracking.
Lowest spot
Price by provider
Cheapest Spot Trainium2 offer at each provider, per GPU-hour. Shorter and greener is cheaper.
- AWS🇺🇸$0.54best
Thin vertical line = median across providers ($0.54).
Every provider we track
Pricing
Size
Host
Provider HQ
1 providers · 1 configurations
Sort
- AWSSpot🇺🇸 HQ in United Statesvia AWS pricingfrom · SpotCheapest$0.54/GPU·hCheapest · Spot
- 16× GPU—$0.54 /GPU·h$8.60 /hSpotus-east-2 (+1 regions)Rent ↗
What can the Trainium2 run?
Memory for the weights plus one 32K-token sequence of context, how many GPUs hold it, and what that replica costs at the cheapest in-stock on-demand rate.
| Model | Params | Memory | GPUs | $/hr | $/mo |
|---|---|---|---|---|---|
gpt-oss-20b OpenAI · 20.9B | 20.9B · 3.6B active | 20 GiB | 1 | — | — |
Gemma 3 27B Google · 27.4B | 27.4B | 28 GiB | 1 | — | — |
Qwen3-32B Alibaba · 32.8B | 32.8B | 39 GiB | 1 | — | — |
Llama 3.3 70B Instruct Meta · 70.6B | 70.6B | 76 GiB | 1 | — | — |
gpt-oss-120b OpenAI · 117B | 117B · 5.1B active | 110 GiB | 2 | — | — |
Qwen3-235B-A22B Alibaba · 235B | 235B · 22B active | 225 GiB | 4 | — | — |
DeepSeek-R1 DeepSeek · 684B | 684B · 37B active | 640 GiB | 8 | — | — |
Kimi K3 Moonshot AI · 2.78T | 2.78T · 104B active | 2.53 TiB | 32 | — | — |
Assumes 90% of each GPU's memory is usable and rounds up to 1, 2, 4, 8… GPUs for tensor parallelism (up to 64). $/hr is the GPU count times the cheapest in-stock on-demand rate per GPU; check that the provider rents instances that large. Serving many requests at once needs more memory for context.
Alternatives to the Trainium2
GPUs with similar compute and memory, and their lowest on-demand price right now.
Trainium2 pricing questions
Is there cheaper spot or interruptible Trainium2 capacity?
Yes. Spot (interruptible) Trainium2 capacity starts at $0.54 per GPU-hour. Spot instances can be preempted at any time, so use them for checkpointed training or batch jobs.
What are the Trainium2's key specs?
96 GB HBM3 with 2,900 GB/s of memory bandwidth, Trainium architecture (2024), Trainium form factor. Dense peak: 0.667 PF FP16 / BF16, 1.3 PF FP8, 0.181 PF FP32.
How current are these prices?
Prices were collected on Nov 9, 2026, 21:02 UTC from provider APIs and public catalogs and refresh about every 2 hours. They are indicative list prices; always confirm on the provider's website before booking.