Home / Compute
Synteq
gpurentalprices.com · 4 records in Compute · JSON
Records 4
- paidHot Aisle
Computeidentity emailno cardevidence documentedstale on 2026-12-04
- paidPrime Intellect Marketplace
Computeidentity emailno cardlimit volume-quotaevidence reportedstale on 2026-11-06
- paidSaladCloud community floor
Computeidentity emailno cardlimit retention-clampedevidence reportedstale on 2026-11-18
- paidSynteq
Computeidentity cardlimit write-gatedevidence reportedstale on 2026-11-18
gpurentalprices.com/providers/hotaisleGPU cloudcompute/hot-aisleJSONcomparereport an error
- Free allowance
- AMD-only GPU host: the pricing page lists none of H100, A100 80 GB, L40S, RTX 4090 or L4.
- Verdict
- paidno free allowance; listed for its price
- Cost
- no price stated
- Identity
- emailno card, no phone
- Binding limit
- noneno named constraint binds
- Evidence
- documented11 sources; a first-party page was read and the sentence quoted
- Freshness
- stale on 2026-12-04checked 2026-11-03 (live); stale on 2026-12-04
| Free tier | AMD-only GPU host: the pricing page lists none of H100, A100 80 GB, L40S, RTX 4090 or L4. MI300X virtual machines (1, 2 and 4 GPUs) are USD 2.99 per GPU-hour and 8x MI300X bare metal is USD 3.39 per GPU-hour, billed by the minute; the page names no pricing tier and states no minimum bill. Credits are added to a team before compute is provisioned. A third-party tracker reports a minimum rental of one GPU for one minute and prepaid credits bought by card through Stripe or in USDT and USDC (checked 2026-10-17). unit hours; metered on wall-clock; none |
|---|
| Price | none stated |
|---|
| Binding | none Nothing published binds above a single GPU for a single minute; the whole cost is prepaid and the smallest unit is one minute. |
|---|
| Identity | email The vendor's quick-start page signs a user in with an email address and a verification code sent to it, then credits are added to a team before compute is provisioned. Prepaid credits rather than a subscription, and a third-party tracker records payment by card or by USDT and USDC, so a stablecoin balance is an alternative to a card. |
|---|
| Egress | not stated |
|---|
| Interface | unknown inferred from the tooling the provider documents; not probed |
|---|
| Tooling | not stated |
|---|
| Composes with | The one-minute minimum and per-minute billing make this the only 192 GB card sold in units a benchmark can afford, which is what a trial run needs before a multi-hour commitment. |
|---|
| Reliability | no reliability event recorded |
|---|
| Notes | Storage price, data-out price and the rule for stopped instances are stated by the tracker as unpublished by the vendor, which matters because a 192 GB card implies a large dataset beside it. |
|---|
| Parsed numbers | |
|---|
Evidence 11
| date | source | what it proved |
|---|
| 2026-10-22 | gpurentalprices.com/providers/hotaisle | quoted: 'Hot Aisle rents 1 GPU model today, from $2.99 an hour (MI300X, on-demand)'; source of the rate given as hotaisle.xyz/pricing |
| 2026-10-22 | gpurentalprices.com/providers/hotaisle | quoted billing terms: 'Billing is per minute', 'Minimum rental time: one GPU for one minute', 'prepaid credits by card (Stripe) or USDT and USDC', each checked on 2026-10-17 |
| 2026-10-22 | gpurentalprices.com/mi300x | MI300X market median across 5 providers is USD 2.99/hr, so Hot Aisle matches the median rather than undercutting it |
| 2026-11-03 | hotaisle.xyz/pricing | the vendor pricing page lists MI300X VMs at USD 2.99 per GPU-hour billed by the minute; its table lists no H100, A100, L40S, RTX 4090 or L4 |
| 2026-11-03 | hotaisle.xyz/pricing | the rate and the per-minute billing are read on the vendor's page |
| 2026-11-03 | hotaisle.xyz/pricing | the vendor pricing page was read on this date |
| 2026-11-03 | hotaisle.xyz/pricing | the rate and the billing granularity are first-party |
| 2026-11-03 | hotaisle.xyz/quick-start | sign-in is by email address and a verification code |
| 2026-11-03 | hotaisle.xyz/pricing | 8x MI300X bare metal is also billed by the minute, at USD 3.39 per GPU-hour |
| 2026-11-03 | hotaisle.xyz/pricing | the 8x bare-metal rate is USD 3.39 per GPU-hour |
| 2026-11-03 | hotaisle.xyz/pricing | the vendor pricing page confirms the MI300X VM rate and per-minute billing |
More fields 1
| facets | {"class": "gpu-broker-rate-floor","free_tier_kind": "none","shape": {"vcpu": null,"ram_gib": null,"gpu_class": "MI300X","gpu_count": 1,"vram_gib": 192}} |
|---|
Unknown values 7
- egress metered: egress metering was not published
- egress terms: a source was read and is silent on this field
- interface: the access interface was not established
- per object cap stated: the field does not apply to this kind of offer
- reliability note: nothing of this kind was observed
- retention stated: the field does not apply to this kind of offer
- tooling: a source was read and is silent on this field
gpurentalprices.com/providers/primeintellectGPU cloudcompute/prime-intellect-marketplaceJSONcomparereport an error
- Free allowance
- Marketplace lane of Prime Intellect, priced per GPU-hour: H200 141GB at USD 3.14, B200 180GB at USD 3.49, B300 288GB at USD 4.99, and spot or interruptible H100 at USD 0.94, which is the cheapest H100 hour recorded anywhere.
- Verdict
- paidno free allowance; listed for its price
- Cost
- no price stated
- Identity
- emailno card, no phone
- Binding limit
- volume-quotaPrepaid credits are the ceiling: the existing record quotes the documentation, credits are deducted every minute while a Pod is active and Pods stop when credits run out, so the balance is what limits a run.
- Evidence
- reported2 sources; a third party said so
- Freshness
- stale on 2026-11-06checked 2026-10-22 (live); stale on 2026-11-06
| Free tier | Marketplace lane of Prime Intellect, priced per GPU-hour: H200 141GB at USD 3.14, B200 180GB at USD 3.49, B300 288GB at USD 4.99, and spot or interruptible H100 at USD 0.94, which is the cheapest H100 hour recorded anywhere. Prices checked 2026-10-22. unit hours; metered on wall-clock; none |
|---|
| Price | none stated |
|---|
| Limits | - volume-quota (binding) Prepaid credits are the ceiling: the existing record quotes the documentation, credits are deducted every minute while a Pod is active and Pods stop when credits run out, so the balance is what limits a run. source
|
|---|
| Identity | email Unchanged from the existing record: the platform uses prepaid credits, and the grant programme that once existed is gone. |
|---|
| Egress | not stated |
|---|
| Interface | unknown inferred from the tooling the provider documents; not probed |
|---|
| Tooling | not stated |
|---|
| Composes with | USD 0.94 for an interruptible H100 hour is half the cheapest on-demand H100 in the same market, so a checkpointing run that tolerates preemption gets two H100 hours for the price of one anywhere else. |
|---|
| Reliability | no reliability event recorded |
|---|
| Notes | Extends the existing prime-intellect.json record, which covers the sandbox pricing only, and prime-intellect-2.json, which records the grant programme as dead. The marketplace rates are third-party reported and the vendor pricing page answered 404 on 2026-10-22, so the rung is reported. A spot rate is not a supply guarantee: the tracker labels it spot, and the existing record's note that Pods stop when credits run out is the practical limit. |
|---|
| Parsed numbers | |
|---|
Evidence 2
| date | source | what it proved |
|---|
| 2026-10-22 | gpurentalprices.com/providers/primeintellect | the tracker names Prime Intellect as the spot floor for H100 at USD 0.94 per GPU-hour and as the on-demand floor for H200 at USD 3.14, B200 at USD 3.49 and B300 at USD 4.99 |
| 2026-10-22 | gpurentalprices.com/ | quoted: 'Spot from $0.94 (Prime Intellect)' and 'Cheapest on-demand H100 today: $1.80/hr at HyperAI' with the H100 median at USD 3.29 |
More fields 1
| facets | {"class": "gpu-marketplace","free_tier_kind": "none","shape": {"vcpu": null,"ram_gib": null,"gpu_class": "H100","gpu_count": 1,"vram_gib": 80}} |
|---|
Unknown values 7
- egress metered: egress metering was not published
- egress terms: a source was read and is silent on this field
- interface: the access interface was not established
- per object cap stated: the field does not apply to this kind of offer
- reliability note: nothing of this kind was observed
- retention stated: the field does not apply to this kind of offer
- tooling: a source was read and is silent on this field
gpurentalprices.com/providers/synteqGPU cloudcompute/synteqJSONcomparereport an error
- Free allowance
- No free tier. The vendor catalogue lists five GPU plans, every one with in_stock false: Intel Arc Pro A40 $45 a month, NVIDIA RTX PRO 4000 $185, RTX PRO 5000 $450, RTX PRO 6000 $885 and B300 $4555.
- Verdict
- paidno free allowance; listed for its price
- Cost
- no price stated
- Identity
- cardcard required, no phone
- Binding limit
- write-gatedA one-month prepaid commitment with no partial refund is the binding term: a cancelled month is a lost month, and none of the five GPU plans in the vendor catalogue was in stock on the day checked.
- Evidence
- reported9 sources; a third party said so
- Freshness
- stale on 2026-11-18checked 2026-11-03 (live); stale on 2026-11-18
| Free tier | no free tier. The vendor catalogue lists five GPU plans, every one with in_stock false: Intel Arc Pro A40 $45 a month, NVIDIA RTX PRO 4000 $185, RTX PRO 5000 $450, RTX PRO 6000 $885 and B300 $4555. H100, A100 80 GB, L40S, RTX 4090 and L4 are not listed. The price index divides $185 a month by 730 hours to USD 0.25 per GPU-hour for the RTX PRO 4000, which is not a published hourly rate. The vendor GPU page lists bare-metal servers from $1545 a month (8 x RTX Pro 4000) and $1785 a month (2 x RTX PRO 6000). Block storage is USD 4 per TB per month, bandwidth is included on every plan, billing is prepaid monthly with a one-month minimum rental, and cancelling gives no partial refund unit hours; metered on wall-clock; none |
|---|
| Price | none stated |
|---|
| Limits | - write-gated (binding) A one-month prepaid commitment with no partial refund is the binding term: a cancelled month is a lost month, and none of the five GPU plans in the vendor catalogue was in stock on the day checked. source
A one-month prepaid commitment with no partial refund is the binding term: a cancelled month is a lost month, and none of the four configurations was in stock on the day checked. |
|---|
| Identity | card The first month is paid up front and the tracker records no free tier. |
|---|
| Egress | 'None of these listed configurations is reported in stock'; and the terms 'Billing is monthly, prepaid', 'Minimum rental time: 1 month; cancelling gives no partial refund', 'Data out: bandwidth included on every plan' |
|---|
| Interface | unknown inferred from the tooling the provider documents; not probed |
|---|
| Tooling | not stated |
|---|
| Composes with | A USD 185 month with bandwidth included is a fixed-cost way to hold one card continuously, but LeaderGPU's USD 140 month for a 3090 is cheaper and this is a commitment, so it only composes with a workload that runs for most of 730 hours. |
|---|
| Reliability | no reliability event recorded |
|---|
| Notes | Supply is the finding and it is explicit: zero in stock. The USD 0.25 per hour is the tracker's division of a monthly price, not a published hourly rate, and a month used for 100 hours costs USD 1.85 per hour on the same invoice. |
|---|
| Parsed numbers | |
|---|
Evidence 9
More fields 1
| facets | {"class": "committed-gpu-host","free_tier_kind": "none","shape": {"vcpu": null,"ram_gib": null,"gpu_class": "RTX PRO 4000 Blackwell","gpu_count": 1,"vram_gib": null}} |
|---|
Unknown values 5
- interface: the access interface was not established
- per object cap stated: the field does not apply to this kind of offer
- reliability note: nothing of this kind was observed
- retention stated: the field does not apply to this kind of offer
- tooling: a source was read and is silent on this field