Tencent CNB

cnb.cool · Compute, Storage · 2 records · JSON

Tencent CNB Computecaveatany-phoneno cardphoneDrequest-rate
CNB 云原生构建 (cnb.cool, Tencent Cloud Native Build, community edition) Storagecaveatnational-idno cardRper-object-cap

Tencent CNB Compute

caveatany-phoneno cardphoneDrequest-rate

docs.cnb.cool/zh/pricing.html · dev environment · compute/tencent-cnb · JSON · compare

Free tier	translated from the CNB pricing table, read 2026-10-22: cloud-native build CPU free quota 160 core-hours a month; cloud-native development CPU free quota 1,600 core-hours a month; cloud-native build GPU free quota 0 core-hours a month; cloud-native development GPU free quota 0 core-hours a month; GPU overage at CNY 0.5 per core-hour; repository storage 100 GiB; object storage 100 GiB; 500 AI credits a month. unit hours; metered on wall-clock; allowance
Price	none stated
Per-object cap	not applicable
Retention	not applicable
Egress	not stated
Identity	any-phone not published
Limits	request-rate 1,600 core-hours a month of development allowance is the only lane left, because the GPU lane went from a quota to zero on 2026-08-01 and a build that needs a card now bills from the first core-hour. source 1,600 core-hours a month of development allowance is the only lane left, because the GPU lane went from a quota to zero on 2026-08-01 and a build that needs a card now bills from the first core-hour.
Binding	request-rate 1,600 core-hours a month of development allowance is the only lane left, because the GPU lane went from a quota to zero on 2026-08-01 and a build that needs a card now bills from the first core-hour.
Interface	unknown inferred from the tooling the provider documents; not probed
Tooling	not stated
Composes with	1,600 CPU core-hours a month runs a continuous build and test loop for one person, and it holds the build step of any chain whose GPU step has moved elsewhere.
Reliability	no reliability event recorded
Notes	This is a correction and a refutation in one record. A post dated 2026-04-27 made a shared L40 at more than 40 GB of VRAM look like free capacity; the vendor's own pricing page read 2026-10-22 lists the GPU free quota as 0. The corpus holds cnb only as a storage record (dossier/storage/cnb.json), so the compute side was never recorded and the rot is invisible there. Identity: a Tencent Cloud account, which for a mainland registration is a real-name check with a phone number as the visible cost.
Evidence rung	documented
Checked	2026-10-22 (live); stale on 2026-10-24

Evidence (4)

date	source	what it proved
2026-10-22	docs.cnb.cool/zh/pricing.html	the pricing table: build CPU 160 core-hours a month free, development CPU 1,600 core-hours a month free, both GPU lines 0 core-hours a month free, storage 100 GiB for repositories and 100 GiB for objects
2026-07-01	cloud.tencent.com/announce/detail/2349	quoted announcement, translated: from 2026-08-01 CNB cancels the free quota for cloud-native build GPU and cloud-native development GPU; GPU moves to pay-as-you-go; organizations without a bound Tencent Cloud budget cannot use GPU at all until they bind one
2026-04-27	linux.do/t/topic/2068176	the pre-withdrawal state, translated: a .cnb.yaml can request the cnb amd64 GPU L40 tag, giving a shared L40 with more than 40 GB of VRAM, with a single session of at most 4 hours
2026-07-26	linux.do/t/topic/2635751	an account's own quota panel showing a development GPU monthly quota of 1,600 core-hours, then a reply noting the announcement cancels it, then an answer on 2026-08-01 saying the GPU still works, which is a rollout lag and not a revocation

More fields (1)

{
 "wave2": {
  "class": "ci-gpu-runner",
  "free_tier_kind": "allowance",
  "shape": {
   "vcpu": 8,
   "ram_gib": 16,
   "gpu_class": null,
   "gpu_count": 0
  }
 }
}

Unknown or withheld values (10)

$.egress.metered egress-unknown
$.egress.terms not-stated
$.identity_cost.detail process
$.interface interface-unknown
$.per_object_cap.notes process
$.per_object_cap.stated not-applicable
$.reliability_note none-observed
$.retention.notes process
$.retention.stated not-applicable
$.tooling not-stated

CNB 云原生构建 (cnb.cool, Tencent Cloud Native Build, community edition) Storage

caveatnational-idno cardRper-object-cap

cnb.cool · paas · storage/cnb · JSON · compare

Free tier	Per top-level org, free tier per month: '仓库存储 100 GiB' + '对象存储 100 GiB' (object storage covers '制品、LFS 对象、图片及附件'), plus 160 build-CPU core-hours and 1600 dev-CPU core-hours. Over-quota is metered at ¥1/GiB/month, and storage is billed on daily average minus the free allowance — so staying under 100+100 GiB per org is genuinely free. unit GiB; metered on stored-bytes
Price	none stated
Per-object cap	256 MB per file (256000000 bytes)
Retention	not stated
Egress	Not metered separately in the pricing table — the only storage line items are 仓库存储 and 对象存储, both by GiB-month; there is no egress/流量 row. Public release-attachment downloads run on the cnb.cool CDN. Treat as unmetered-until-proven; I did not find an egress line to quote.
Identity	national-id national-id (WeChat login / 腾讯云 account; the docs' usage-management flow binds a Tencent Cloud budget). Not anonymous.
Limits	per-object-cap 256 MB per file source per-object-cap: 64 GB per release/commit attachment (provider's own plugin README: '目前暂支持 64GB 以内的文件上传'). repo-blob cap: 256 MB per file in the git tree, with the maintainers explicitly steering large files to release attachments ('此256M之限制...当以附件方式为之'). request-rate: none documented. Billing: over the free 100 GiB the org is charged ¥1/GiB/month at month start — silent overage, not a hard stop.
Binding	per-object-cap the per-object cap sets the shard size and therefore the object count; it bounds the total more tightly than any headline quota
Interface	rest, webdav, webdav-proxy, cli rest
Tooling	OpenList ships a first-class `cnb_releases` driver (github.com/OpenListTeam/OpenList, drivers/cnb_releases — fields: repo, token, use_tag_name, default_branch), i.e. a maintained WebDAV/HTTP front end over CNB release attachments. Also the official `cnbcool/attachments` plugin ('支持 commit 和 release 上传或下载附件，目前暂支持 64GB 以内的文件上传') and a Rust CLI (`cnb release download/--share`).
Composes with	Free allowance is per top-level org, and '每个顶级组织独立计费' — so N orgs = N x 200 GiB. Two independent multipliers: (a) unlimited repos inside one org, each with unlimited releases, each attachment up to 64 GB; (b) unlimited orgs at 200 GiB each. 10 TiB hot = ~100 orgs at the free ceiling, or 160 release attachments at 64 GB in a single org. Per-file ceiling (64 GB) is 32x GitHub Releases' 2 GiB.
Reliability	Community edition launched public beta 2025-08; beta-free extras expired 2025-08-31, so the 100+100 GiB figures are the steady state. No ban/wipe reports found; overage is billed rather than blocked, which is the real hazard.
Notes	Per top-level org, free tier per month: '仓库存储 100 GiB' + '对象存储 100 GiB' (object storage covers '制品、LFS 对象、图片及附件'), plus 160 build-CPU core-hours and 1600 dev-CPU core-hours. Over-quota is metered at ¥1/GiB/month, and storage is billed on daily average minus the free allowance — so staying under 100+100 GiB per org is genuinely free.
Evidence rung	reported One page read moves it to documented.
Checked	2026-10-21 (inherited); stale on 2026-11-04
Parsed numbers	`$.limits[0].value` limit:per-object-cap 256.0 MB

Evidence (3)

date	source	what it proved
2026-10-21	docs.cnb.cool/zh/pricing.html	'仓库存储 \| 100 GiB \| ¥1/GiB/月 \| Git 对象' and '对象存储 \| 100 GiB \| ¥1/GiB/月 \| 制品、LFS 对象等'; '每个顶级组织独立结算'
2026-10-21	cnb.cool/wmde/attachments	'支持 commit 和 release 上传或下载附件，目前暂支持 64GB 以内的文件上传'
2026-10-21	github.com/OpenListTeam/OpenList/tree/main/drivers/cnb_relea	maintained driver with release-attachment upload/download, proving a working non-web client

Unknown or withheld values (4)

$.free_tier.notes process
$.per_object_cap.notes not-extracted
$.retention.notes not-extracted
$.retention.stated not-extracted