GPU MODELS / TESLA T4
Tesla T4 cloud GPU pricing
1 listings across 1 providers. Cheapest published rate: $0.134/GPU/hr at Vast.ai. Sorted by per-GPU hourly price. Every row links to the provider's own published price.
WHAT IS THE TESLA T4
The T4 is NVIDIA's low-power inference workhorse: a 70-watt single-slot card that became the default cost floor for model serving. It has no NVLink and modest memory bandwidth, but its power draw and price made it the most widely deployed inference GPU in cloud catalogs. In 2026 it is a legacy part - still the cheapest way to serve small models, but outclassed by the L4 on every metric.
TENSOR COMPUTE
65 TFLOPS FP16 tensor (130 TOPS INT8)
WHAT IT IS GOOD AT
Small-model inference
The design target. A 7B model at INT8 or FP16 fits comfortably in 16 GB, and 65 TFLOPS of FP16 is enough for moderate request rates.
Video transcoding
Strong fit - Turing's NVENC/NVDEC blocks handle multi-stream transcode at very low power.
Fine-tuning
Workable for LoRA on small models only. 16 GB and 320 GB/s rule out anything serious.
WHO SHOULD NOT RENT THE TESLA T4
Anything beyond small models: 16 GB and 320 GB/s rule out 13B+ serving and all serious training.
New deployments where the L4 rents within ~30% of T4 prices - the successor is faster at the same power.
| Provider | Config | VRAM | Location | $/hr | $/GPU/hr | Source |
|---|
 Vast.ai | Tesla T4 × 2 | 15 GB | Pennsylvania, US | $0.269 | $0.134 | receipt |
THE TESLA T4 MARKET, BY THE NUMBERS
1 providers publish 1 Tesla T4 configurations on the panel today. Published per-GPU hourly rates run from $0.134 (Vast.ai) to $0.13 (Vast.ai), a 1x spread between the cheapest and the most expensive published rate for the same chip. The median listing sits at $0.13/GPU/hr. 1 rows were reported in stock by the provider at last observation; the rest are published catalog rates. All rates are on-demand - committed-use and reserved pricing is excluded.
TESLA T4 PRICE TREND, DAILY PANEL SNAPSHOTS
Across every Tesla T4 row on the panel, the daily floor held flat at $0.134/GPU/hr over 2 days of tracking. Snapshots run daily; the full history is public in the repo.
TESLA T4 PROVIDER NOTES
Vast.ai · 1 config · 1 region · 1 in stockSets the panel floor for this model.
from $0.134/hrBUYING TESLA T4: WHAT THE PANEL SAYS
At the current floor, one Tesla T4 running around the clock costs about $98 a month (730 hours of on-demand arithmetic). Compare the alternatives below before committing - the same budget often buys more than one chip class.
TESLA T4 VS THE ALTERNATIVES
The L4 is the T4's direct successor: roughly 2x the FP16 throughput and 2.5x the INT8, newer video engines, at a similar 72 W. Unless the T4 is dramatically cheaper on the panel, the L4 is the better default.
The A10 adds 8 GB more VRAM and nearly double the compute at 150 W. Pick it when a model does not fit in 16 GB or request rates saturate the T4.
Compare live rates:
TESLA T4 PRICING FAQ
What is the cheapest Tesla T4 cloud GPU right now?
The cheapest published Tesla T4 rate on the panel is $0.134 per GPU-hour at Vast.ai (Tesla T4 x 2, Pennsylvania, US). Every price links to the provider's own published rate.
How much does a Tesla T4 cost per hour?
Across 1 providers tracked today, published per-GPU hourly rates for Tesla T4 run from $0.13 to $0.13, with a median of $0.13. Committed-use and reserved rates are excluded; everything shown is on-demand.
How many providers rent Tesla T4 GPUs?
1 providers publish 1 distinct Tesla T4 configurations on Compute Cafe. 1 of those rows were reported in stock by the provider's live endpoint at last observation.
Are Tesla T4 prices going up or down?
Over the 2 days Compute Cafe has tracked the panel, the Tesla T4 floor is holding flat: $0.134 to $0.134/GPU/hr (+0.0%). Daily snapshots are public in the repo, so the trend is auditable.
What does a Tesla T4 cost per month?
At the current floor of $0.134/GPU/hr, a Tesla T4 running 24/7 costs about $98 per month (730 hours of on-demand arithmetic on the cheapest published rate). Committed-use contracts can price lower; the panel tracks on-demand rates only.
Is the T4 still worth renting?
Only when it is the cheapest option for a small workload. The panel often shows T4 rates under $0.20/GPU/hr; at that price it is fine for dev work, small-model serving and video pipelines. For anything bigger, an L4 at a small premium is usually the better deal.
Can a T4 run LLM inference?
Yes for small models: a quantized 7B model runs well, a 13B fits tightly at INT8. Larger models do not fit in 16 GB without splitting across cards, and the T4 has no fast GPU-to-GPU interconnect, so multi-card inference is slow.
See also alternatives to the Tesla T4 · cheapest GPU-hour per model · inference chips · training chips · by location