fshy

Home

BytePlus AI

byteplus.com · 2 records in Compute, Inference · JSON

Records 2

ai.byteplus.com/ark/region:ap-southeast-1/docs/free-inferencInference API, one time free tokensinference/byteplus-modelark-free-tokensJSONcomparereport an error

Free allowance
A one-time free inference quota per model, 500K tokens per model, granted to new users on registration.
Verdict
triala time-boxed or one-time trial, not a recurring allowance
Cost
free tier
Identity
emailcard unknown, phone unknown
Binding limit
volume-quota500K tokens per model, one-time; offsets online pay-as-you-go inference only, not batch inference, plugins, knowledge bases or context-cache storage
Evidence
documented30 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 tierA one-time free inference quota per model, 500K tokens per model, granted to new users on registration. It is calculated separately for each model and shared under the primary account, offsets pay-as-you-go online inference tokens only, and in Free Tokens Only mode the service is suspended near the quota so no fee accrues. The activity page states the quota does not expire but is subject to change
unit tokens; metered on request-count; one-time-quota
Pricenone stated
Limits
  • volume-quota (binding) 500K tokens per model, one-time; offsets online pay-as-you-go inference only, not batch inference, plugins, knowledge bases or context-cache storage source
  • request-rate Model-list maximums per primary account (theoretical, subject to platform load; all endpoints of a base model share them), for the models the pricing page lists at 500k tokens: Seed 2.0 lite, mini, code-preview and pro 30K RPM and 1500K TPM; DeepSeek V4 pro and flash 260425 versions 15K RPM and 1500K TPM; DeepSeek V4 pro-ga and flash-ga 0.5K RPM and 1000K TPM; Seed 2.1 turbo 0.5K RPM and 1000K TPM (its free-quota row reads 500 tokens); retiring Seed 1.8 30K RPM and 1500K TPM, Seed 1.6 and Seed 1.6 flash 15K RPM and 800K TPM, DeepSeek V3.2 15K RPM and 1500K TPM. Skylark-pro has no row on the model list; which model version carries the quota is not stated source
Bindingvolume-quota the per-model 500K token quota is the only stated cap
Identityemail The sign-up form asks for email, username and password, or Google or GitHub sign-in. The quota page says to register and authenticate as a BytePlus user; the help centre says individual accounts reach a limited subset of products, business accounts need a business license and company details, and a valid credit card may be required to activate pay-as-you-go models; Free Tokens Only mode is open to identity-verified personal accounts and enterprise accounts; the campaign terms require enterprise information submission and judge uniqueness from email verification, ID documents, domain names, mobile numbers, devices and IP addresses. unverified: whether a card, phone number or identity verification is required before the quota is granted, which only an account shows
Egressnot published
Interfacerest
Toolingpart of the text of Limits, above
Composes witheach model carries its own quota, so one account holds several separate pools; Free Tokens Only mode stops service at the quota and cannot accrue a charge
Reliabilitya separate Data Collaboration Rewards Campaign advertising 5M tokens daily per model behind real-name verification: the campaign page calls it an ongoing benefit while the activity page carries an earlier stated end date; daily resource pack distribution for 18 models, among them Seed 2.0 pro, lite and mini, Seed 1.6, 1.8, DeepSeek V3.2 and V4 pro and flash GA, ended on October 8, 2026
Notesthe quota is per model and per primary account; the free-trial page lists DeepSeek V3.1 and GPT-OSS-120B among the 500K-token models
Evidence notedocumented quota size and scope are documented; the identity requirement is unverified beyond the quoted terms
Parsed numbers
  • has-free-tier 1
  • request-rate 500 tokens
Evidence 30
datesourcewhat it proved
2026-10-28ai.byteplus.com/ark/region:ap-southeast-1/docs/free-inferencnew users receive a free inference trial quota
2026-10-28ai.byteplus.com/ark/region:ap-southeast-1/docs/free-inferencFree Tokens Only mode consumes only the 500k token platform quota
2026-10-28ai.byteplus.com/ark/region:ap-southeast-1/docs/free-inferencthe quota is per model and shared under the primary account
2026-10-28www.byteplus.com/en/activity/freethe free-trial page lists 500K free tokens per model for DeepSeek V3.1 and GPT-OSS-120B
2026-10-28www.byteplus.com/en/activity/freethe activity page states the free quota does not expire but can change
2026-10-28docs.byteplus.com/en/docs/legal/termsandconditions_modelark_the free-token campaign terms require account registration and enterprise information submission
2026-10-28www.byteplus.com/en/activity/freea separate daily-token campaign behind corporate verification had a stated end date of March 31, 2026
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/model-listseed-2-0-lite lists Max RPM 30K and Max TPM 1500K
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/model-listdeepseek-v4-pro 260425 lists 15K RPM and 1500K TPM
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/model-listdeepseek-v4-pro-ga lists 0.5K RPM and 1000K TPM
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/model-listdola-seed-2-1-turbo lists 0.5K RPM and 1000K TPM
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/model-listretiring seed-1-6 lists 15K RPM (800K TPM on the next row)
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/model-listretiring deepseek-v3-2 lists 15K RPM (1500K TPM on the next row)
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/model-listthe listed rates are theoretical maxima that vary with platform load
2026-11-03www.byteplus.com/en/product/modelark/pricingthe pricing page lists DeepSeek-V4-pro at 500k free tokens
2026-11-03www.byteplus.com/en/product/modelark/pricingthe pricing page lists Dola-Seed-2.1-turbo free quota as 500 tokens
2026-11-03www.byteplus.com/en/product/modelark/pricingSkylark-pro is listed at 500k free tokens
2026-11-03console.byteplus.com/auth/signupthe sign-up form lists email, username and password fields
2026-11-03console.byteplus.com/auth/signupthe sign-up form offers Google sign-in
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/free-inferencthe quota page says to register and authenticate as a BytePlus user
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/free-tokens-oFree Tokens Only mode is open to identity-verified personal and enterprise accounts
2026-11-03ai.byteplus.com/en/help/article/what-identity-verification-rbusiness accounts need a business license and company details
2026-11-03ai.byteplus.com/en/help/article/what-is-the-difference-betweindividual accounts reach a limited subset of products
2026-11-03ai.byteplus.com/en/help/article/why-cant-i-bind-my-credit-caa valid credit card may be required to activate pay-as-you-go models
2026-11-03docs.byteplus.com/en/docs/legal/termsandconditions_modelark_the campaign terms require enterprise information submission
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/collaborationdaily distribution for 18 models ended on October 8, 2026
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/collaborationthe campaign page calls the 5M daily tokens an ongoing benefit
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/collaborationthe campaign requires real-name authentication
2026-11-03www.byteplus.com/en/activity/freethe activity page still states March 31, 2026 for the campaign
2026-11-03ai.byteplus.com/ark/region:ap-southeast-1/docs/free-inferencthe quota is per model and shared under the primary account
Unknown values 3
  • egress metered: egress metering was not published
  • per object cap stated: the field does not apply to this kind of offer
  • retention stated: the field does not apply to this kind of offer

www.byteplus.com/en/solutions/ai-cloud-native-sandboxContainer and VM sandboxcompute/byteplus-ai-cloudJSONcomparereport an error

Free allowance
Serverless microVM sandboxes that the vendor says start in under 150 ms, scale to tens of thousands created concurrently and to millions of vCPUs, support stateful long-running sandboxes, ship preconfigured templates, and bill at millisecond granularity with no charge while nothing is executing.
Verdict
paidno free allowance; listed for its price
Cost
no price stated
Identity
email?placeholder, not evidence; card unknown, phone unknown
Binding limit
write-gatedThe rate card and the signup gate are both behind an account: the public documentation renders as a script shell, so the only published numbers are third-party, and whether a personal account can create a sandbox at all was not established.
Evidence
reported3 sources; a third party said so
Freshness
stale on 2026-11-06checked 2026-10-22 (live); stale on 2026-11-06
Free tierServerless microVM sandboxes that the vendor says start in under 150 ms, scale to tens of thousands created concurrently and to millions of vCPUs, support stateful long-running sandboxes, ship preconfigured templates, and bill at millisecond granularity with no charge while nothing is executing. Templates, large-scale image distribution, arbitrary specifications and MCP, API and SDK access are listed. No free tier is published.
unit hours; metered on wall-clock; none
Pricenone stated
Limits
  • write-gated (binding) The rate card and the signup gate are both behind an account: the public documentation renders as a script shell, so the only published numbers are third-party, and whether a personal account can create a sandbox at all was not established. source
Identityemail (placeholder) unverified: no page stating the signup requirement was read; the record's evidence covers the sandbox mechanics (sub-150 ms starts, millisecond billing with no charge while idle, MCP/API/SDK access), a third-party rate of USD 0.0351 per vCPU-hour and USD 0.0081 per GiB-hour, and a BytePlus billing page that returned only its navigation shell. Whether a personal account can create a sandbox at all was not established.
Egressnot stated
Interfacerest inferred from the tooling the provider documents; not probed
Toolingnot stated
Composes withA microVM that bills only while executing is the right shape for evaluation fan-out, where a thousand short sandboxes each live a few seconds; it is the wrong shape for a service that must stay up, because the published selling point is that idle costs nothing precisely by not being a persistent host.
Reliabilityno reliability event recorded
NotesRoutes tried: the product solution page (read, product facts only), docs.byteplus.com/en/docs/faas/Billing_overview (navigation shell, no table), docs.byteplus.com/en/docs/faas/Overview_of_cloud_sandbox (same shell), and a broad search that returned only third-party comparison tables. Identity is left null rather than guessed: BytePlus sells to enterprises, and whether a personal account can top up without a company was not established. This is the one provider where the door is more likely to be the finding than the price.
Parsed numbers
  • has-free-tier 0
Evidence 3
datesourcewhat it proved
2026-10-22www.byteplus.com/en/solutions/ai-cloud-native-sandbox'sandboxes can start within 150ms and can be created concurrently at a scale of tens of thousands'; 'up to millions of vCPUs'; 'millisecond-level billing with no charges when not executing'; each sandbox is a microVM; MCP, API and SDK access
2026-10-22boxyard.dev/providers/byteplus-cloud-sandboxa third-party comparison lists USD 0.0351 per vCPU-hour and USD 0.0081 per GiB-hour with no free tier, which is a claim about a third party and not a reading of BytePlus's own rate card
2026-10-22docs.byteplus.com/en/docs/faas/Billing_overviewthe billing overview page returns only its navigation shell with no rate table, so the vendor's own numbers could not be read
More fields 1
facets{"class": "regional-sandbox-unpriced","free_tier_kind": "none","shape": {"vcpu": null,"ram_gib": null,"disk_gib": null,"gpu_class": null,"gpu_count": null}}
Unknown values 6
  • egress metered: egress metering was not published
  • egress terms: a source was read and is silent on this field
  • 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