APIs, hosting & data

Catch: Requests fail with an error past the daily cap (resets 00:00 UTC); some models need a paid billing method.

Catch: Trial keys are rate limited and may not be used for production or commercial purposes.

Catch: Free-tier content is used to improve Google products; newest models such as Gemini 4 Argon are not available.

Catch: Low per-model rate limits.

Catch: Free accounts get no included Inference Providers credits; you buy credits or upgrade to PRO ($9/month with $2 of credits a month).

Catch: US-hosted cloud models and fuller web search start on Bionic+ ($20/month); local model speed depends on your hardware.

Catch: Only 1,000 retrievals a month and one project; graph memory is a Pro feature.

Catch: Free plan Studio and API limits are not published; Pro adds monthly API credits and higher Vibe limits.

Catch: Not for production use; production requires an NVIDIA AI Enterprise license (from $4,500 per GPU per year). Some free model endpoints are deprecated at short notice (for example GLM 5.3 after October 28, 2026), so check the model's page.

Catch: Starter models only; more cloud models need added credits.

Catch: Capped at 50 requests a day, with no budgets, spend controls or prompt caching on Free.

Catch: Anonymous use is capped at 2 requests per minute per model, and requests over the limit get a 429 error. The 400-requests-per-minute keyed tier needs a Public Cloud project.

Catch: Beta with unpublished rate limits and no end date set; usage data may be used to improve the models. Japanese-focused models.

Catch: The Basic workload allowance is not quantified, and requests are blocked once it runs out; Free cannot buy overage.

Catch: The free tokens are a one-time allotment per account, not monthly; Batch API usage is billed and older models get no free tokens.

Catch: Only 50 build minutes and 1 GB of LFS storage, shared by the whole workspace.

Catch: Builds run one at a time on Free and time out after 20 minutes.

Catch: Durable Objects and higher limits need the $5/month Paid plan.

Catch: Meant for personal use and smaller projects; wildcard subdomains are on paid plans.

Catch: Actions minutes on private repos are capped at 2,000 a month.

Catch: Not allowed for online businesses or e-commerce; private-repo Pages need a paid plan.

Catch: AI features require buying GitLab Credits.

Catch: Your site must get a few visits every month to count as active. InfinityFree does not sell domain names, so a custom domain has to come from elsewhere.

Catch: Persistent servers and dedicated PostgreSQL are on paid plans.

Catch: Usage beyond the $30 monthly credit is billed as compute, and egress past 1 TiB is $0.04 per GiB (from October 1, 2026); Team ($250/mo) is needed for more seats, containers and credits.

Catch: Custom domains, SFTP and WebDAV uploads, more than one site and some file types need Supporter ($5/month).

Catch: 300 credit monthly limit.

Catch: Usage beyond the Sandbox allowance is pay as you go; Northflank says entering a card only verifies it.

Catch: Oracle places temporary authorization holds on the card (no charges), Always Free instances must be in your home region, and Oracle notes you may hit out-of-host-capacity errors.

Catch: Your code can only reach specific websites over HTTP(S), and the CPU allowance is 100 seconds (Developer gets 5,000 a day). Scheduled tasks and always-on tasks start on Developer ($10/month).

Catch: After the trial, $1 a month covers only very small apps; the Hobby plan has a $5 monthly minimum.

Catch: Free web services spin down after 15 minutes without traffic, and free Postgres databases expire 30 days after creation.

Catch: Each site can be at most 1GB with 1 build at a time. Usage past the limits is billed.

Catch: Each run stops after 1 minute, logs are kept for only 3 days and you can only email yourself. Custom domains, 1-minute cron and AI (Townie) credit start on Pro ($21/month).

Catch: Hobby is for personal, non-commercial use only.

Catch: Only 1 GB of storage and RAM, a single backup, and Aiven may shut down services left unused for a long time.

Catch: Free projects pause after 1 week of inactivity, and you're limited to 2 projects.

Catch: Limits reset daily at 00:00 UTC; once you hit them, queries fail until the reset.

Catch: Past the included amounts you pay as you go (for example $2.20 per extra 1M function calls).

Catch: Some services and higher usage need the Blaze pay-as-you-go plan.

Catch: Meant for learning and exploring; 512 MB cap.

Catch: Compute is capped at 2 CU (8 GB RAM).

Catch: Only one project, and it is paused after a week of inactivity.

Catch: Usage limits apply; community support only.

Catch: Past 200k operations or 2 GB you need the Starter plan ($10/month).

Catch: Meant for testing and prototypes: one small node without high availability; growing past 1 GB RAM or 4 GB disk means moving to the paid Standard Tier.

Catch: Only 30 MB in a single database, with a best-effort SLA and community support only.

Catch: Projects pause after 1 week of inactivity; limit of 2 active projects.

Catch: Capped at 1,000 API requests a day on 0.25 vCPUs, with community support only.

Catch: Point-in-time restore limited to 1 day; community support only.

Catch: Single database with 500K command cap.

Catch: Only 1 collection (up to 3 tenants) and 1,000 Query Agent requests a month, with basic support.

Catch: Files are capped at 250 MB each and only 1 version is kept; Personal Pro ($10 a month billed annually) raises this to 100 GB of storage and 5 GB uploads.

Catch: The free tier covers Standard storage only.

Catch: 25 credits are shared across storage, bandwidth and transformations, and custom domains are paid.

Catch: Only 2 GB.

Catch: 15 GB is shared across Gmail, Drive and Photos.

Catch: Transfer is limited; storage amount is not shown in the page text.

Catch: Only 5 GB; Drive Plus ($4.99 a month) adds 200 GB and 10 years of file version recovery.