# Thunder Compute

> thundercompute.com | Compute; 1 record: free tier, limits, identity cost and evidence.

HTML https://fshy.ajam.dev/p/thundercompute/ | JSON https://fshy.ajam.dev/api/v1/entity/thundercompute.json | how to read a record: https://fshy.ajam.dev/llms-full.txt

## Thunder Compute

- `compute/thunder-compute` | fails-identity | evidence documented | binding write-gated | category GPU cloud
- needs card no, phone no, ID no, institution yes | verified 2026-11-03T08:50:54.214Z (live) | stale on 2026-12-04
- url https://www.thundercompute.com/pricing
- free tier: no free tier: on-demand, non-preemptible GPU instances billed per minute while running, A100 80GB $1.09/hr, H100 SXM $3.20/hr, L40 $0.79/hr and RTX A6000 $0.35/hr (sandboxes bill per second); no L40S, L4 or GeForce card is listed, no minimum bill is stated, the first 100 GB of instance disk per GPU is included and there are no egress fees | unit: none | metered on: none | read on: 2026-10-22 | source: [1]
- per-object cap: not stated | source: [1]
- retention: not stated | source: [1] | note: not stated
- egress: not stated | metered: no | source: [1]
- identity: institution | detail: a university email from a US institution; the page states only students enrolled at US universities qualify | country: US
- interface: unknown
- limits:
  - request-rate: instances: $1.09/hr for an A100 80GB and $3.20/hr for an H100 SXM, billed per minute while running; sandboxes: $0.000819 a GPU-second for an H100 SXM and $0.000542 for an A100 80GB [1]
  - per-object-cap: the first 100 GB of instance disk per GPU is included and bills only while running; snapshots are $0.05 a GB-month [1]
  - per-ip-shaping: data egress is $0.00 a GB [1]
- binding note: the US university enrolment check fails first: there is no other route to the USD 20 credit
- tooling: not stated
- composes with: the same as tooling
- reliability: no reliability event recorded
- notes: Paid-only read only the pricing page, which lists no signup credit, and missed the students page, which publishes USD 20 of credit for A6000 or L40 use. The gate is a US university enrolment rather than a card, so the capacity exists and fails identity.
- evidence note: the pricing page was read on 2026-10-22 and publishes per-second rates per resource, the included disk and the zero egress fee. There is no free tier, so the verdict moves to paid-only
- extras:
  - keep_rate_basis: no field in the source
  - billing_model: unclassified
  - shape.vcpu: 6
  - shape.ram: 48
  - shape.disk: 100
  - shape.gpu_class: RTX A6000
  - shape.gpu_count: 1
  - shape.note: 48 GB of VRAM, 6 vCPU included and up to 8, 48-64 GB of RAM and 100-500 GB of disk
  - session_cap.restart_required: no
  - session_cap.note: instances bill per minute and sandboxes per second, only while running; the page states that instances and sandboxes are never preempted and states no reap and no minimum bill
  - free_tier_kind: none
  - ps.gib: 100
  - ps.price: $0.219 a GB-month past the first 100 GB
  - ps.durability: instance disk and snapshots
  - ps.on_stop: yes
  - ps.on_destroy: no
  - ps.retention: until deleted
  - net.out: yes
  - net.in: yes
  - net.ports: any
  - net.static_ip: yes
  - net.allowance: no egress fees
- evidence:
  - [1] 2026-10-02: 
  - [1] 2026-10-22: 'Billed per second for each resource' with '| RTX A6000 | 48 GB | 6 included · 8 max | $0.35/hr |', 'H100 SXM 80 GB $3.20/hr', 'Instance disk: First 100 GB per GPU included; billed only while running, $0.219/GB/mo', and 'Data egress: No egress fees, $0.00/GB', quoted
  - [2] 2026-10-22: 'Only students enrolled at US universities qualify. Sign up with your university email; credit is added automatically.' and 'Student credit $20.00', quoted
  - [1] 2026-11-03: instances bill per minute while running; the sandbox per-second rate is a separate section of the page | capture: sha256:69cb665e7f187d4e290d949add24477a502f78981bf40409da686e114572c2cf | quote: Instances, including extra vCPUs and disk, are billed per minute only while running.
  - [1] 2026-11-03: the per-second H100 SXM rate belongs to the sandbox table | capture: sha256:69cb665e7f187d4e290d949add24477a502f78981bf40409da686e114572c2cf | quote: | H100 SXM | $0.000819/GPU-sec |
  - [1] 2026-11-03: no preemption is stated | capture: sha256:69cb665e7f187d4e290d949add24477a502f78981bf40409da686e114572c2cf | quote: No. Instances and sandboxes are never preempted. They run until you stop them or, for sandboxes, their TTL ends.
  - [1] 2026-11-03: A100 80GB is listed at $1.09/hr with 80 GB of VRAM | capture: sha256:69cb665e7f187d4e290d949add24477a502f78981bf40409da686e114572c2cf | quote: | A100 80GB | 80 GB | 8 included · 16 max | 64–⁠128 GB | 100–⁠500 GB | $1.09/hr
  - [1] 2026-11-03: H100 SXM is listed at $3.20/hr | capture: sha256:69cb665e7f187d4e290d949add24477a502f78981bf40409da686e114572c2cf | quote: | H100 SXM | 80 GB | 4 included · 16 max | 32–⁠128 GB | 100–⁠500 GB | $3.20/hr
  - [1] 2026-11-03: the listed rates are the on-demand rates; reserved terms are quoted by sales | capture: sha256:69cb665e7f187d4e290d949add24477a502f78981bf40409da686e114572c2cf | quote: Teams can reserve dedicated GPUs for a fixed term at a lower rate than on-demand
- sources:
  - [1] https://www.thundercompute.com/pricing
  - [2] https://www.thundercompute.com/students