# TensorPool

Official pricing: https://tensorpool.dev  
Category: gpu-cloud · Isolation: not published · DISCONTINUED

## Pricing regimes (raw)


## Features

Yes: volumes, ≥24 h sessions, GPU, firewall inside (nftables), extra volumes, shared volumes

No: automatic snapshots, snapshots on demand, agent harness API

Unknown: everything else. Evidence (source + quote) per feature: https://battleships.dev/data/providers/tensorpool.json → feature_evidence

## Caveats

- Discontinued or closed to new customers, confirmed on the live site 2026-09-29. The pricing page (rendered 2026-09-29) still shows the same numbers as the card: B200 SXM $4.99/h, H200 SXM $2.99, B300 SXM $5.49, H100 SXM $1.99, L40S $1.49, CPU $0.015/h. Shared storage $100/TB-mo; Object storage $50/TB-mo + $0.005/1,000 requests, 'No egress or ingress charges'. Standard plan: on-
- On-demand GPU hours: H100 SXM $1.99, H200 $2.99, B200 $4.99, B300 $5.49, L40S $1.49; CPU $0.015/h unit ambiguous on website. Shared storage $100/TB-mo, object $50/TB-mo+$0.005/1000 requests; no ingress/egress fee for object storage. Reservations/enterprise quoted. Conditional $20 GitHub-star credit not universal.
- Null values mean unknown, not free/unlimited. No model-token cost, tax or unverified ancillary charge included.
- Native browser/device/task/credit prices with unknown CPU/RAM are deliberately non-priceable. Check regime before interpreting any partial resource estimate.
- 2026-09-29: API backend engine.tensorpool.dev returns 503 'TensorPool is no longer accepting API requests.' The docs.tensorpool.dev subdomain no longer resolves (no DNS record). The YC directory lists status 'Acquired' (acquirer not named). The last blog post is 2026-03-31, the last CLI commit 2026-04-13 and the last PyPI release 0.0.9 on 2026-03-29. The pricing page is still live but no longer actionable.

## How this provider charges


YC: Winter 2025; directory status **Acquired**. Research scope: public-product-researched. Source: https://www.ycombinator.com/companies/tensorpool

GPU training clusters

## Regime table

| Regime | When it applies | How billed | Numbers | Source |
|---|---|---|---|---|
| Native product; see regime notes | Incomplete or non-resource economics: rates remain null, never interpreted as free. | resource; CPU alloc, memory alloc | Unknown or native-unit billing; see facts below | https://tensorpool.dev |
| Current applicable tariff unverified | Plan limits apply | Monthly fee; credits only as stated | fee=None; included usage=$None; concurrent=None; max session h=None | https://tensorpool.dev |
| Retained disk / GiB-month | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://tensorpool.dev |
| Snapshots / GiB-month | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://tensorpool.dev |
| Egress / GiB | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://tensorpool.dev |
| IPv4 / month | Lifecycle/scope must be checked | Native add-on | unverified, not $0 | https://tensorpool.dev |

## Verified billing facts and gotchas

1. On-demand GPU hours: H100 SXM $1.99, H200 $2.99, B200 $4.99, B300 $5.49, L40S $1.49; CPU $0.015/h unit ambiguous on website. Shared storage $100/TB-mo, object $50/TB-mo+$0.005/1000 requests; no ingress/egress fee for object storage. Reservations/enterprise quoted. Conditional $20 GitHub-star credit not universal.

## Worked example

Requested: 4 vCPU/8 GiB, 50 simultaneous × 8 h/day × 22 days = **8,800 instance-hours**, 30% CPU; 50 GiB snapshots and 100 GiB egress. A month is 730 hours only when explicitly used in rate conversion.

**Total: unknown.** This product does not publish a verified complete 4-vCPU/8-GiB tariff and/or the service bills a different unit. Fifty concurrent instances and eight-hour sessions are not automatically supported. Neither 30% utilization nor retained snapshots can be converted into free resources. Use the native plan terms above and request the missing CPU/RAM, retention, egress and concurrency quote.

## Sources and coverage

- https://www.ycombinator.com/companies/tensorpool
- https://tensorpool.dev/pricing

Page captures and documentation indexes/bundles are in `raw/`; provider JSON lists the evidence paths. HN query results were captured separately as `<id>--hn.txt`; they are discovery leads, not current tariff authority. Downloaded docs do not imply every feature was verified; unsupported assertions remain null.