AI subscription directory
AI subscriptions compared
Choose the LLMs you want to use. Compare matching plans by price, hosting location, and how they handle your data.
- Plans
- 343
- Providers
- 64
- LLM model entries
- 964
Loading the full directory.
Preview: 24 of 343 plans. Loading the full directory.
Command CodeCommand Code Go
Coding
$1.00
Value factor 10.0 ($10 credit)
Command Code Go costs $1 per month and includes $10 of usage in one shared monthly pool. Per-model allocations depend on the model. The official 5-hour and weekly limits conflict, so no single cap is established.
Storage
Unknown
Model processing
US, EU on demand; open source US/EU/Singapore
Training depends on conditions
DPA unknown
GoogleGoogle AI Plus
Chat
$4.99
Entry plan with 2x Gemini limits, 128k context, 200 monthly Flow credits for video and 400 GB storage.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA unknown
Poe (Quora)Poe Basic (10.000 Punkte/Tag)
Chat
$4.99or $4.17/month billed annually
$0.0000166 per credit
Entry Poe plan with 10,000 points per day for GPT-5.5, Claude-Opus-4.7, Gemini and image and video bots.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA unknown
MetaMuse Code Everyday Usage
Coding
$5.00
Flat-rate plan for Meta's terminal coding agent Muse Code with 10 to 50 prompts every 5 hours.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA available
Mistral AIMistral Vibe Pro Education
Chat
$5.99
Value factor 2.5 ($15 credit)
Pro features for verified students at $5.99 per month, with $15 in API credits.
Storage
EU (default hosting)
Model processing
Unknown
Training can be disabled
DPA available
Alibaba Cloud (Qwen)Token Plan Individual Lite
Coding
Limited-time promotional price, list price $8 per month$6.00
Qwen Token Plan Lite: 11,500 credits per month for Qwen3.8, GLM-5.3, DeepSeek-V4 and image models, 1-2 agents at once.
Storage
Singapore
Model processing
Global inference mode
Training depends on conditions
DPA available
ElevenLabsElevenLabs Starter
Audio
Promotional price through November 1, 2026. The first month costs US$1 instead of the regular US$6 price. Prices exclude tax.$6.00or $5.00/month billed annually
$0.20 per minute
30,000 credits a month, about 30 minutes of speech, commercial license and instant voice cloning.
Storage
United States
Model processing
Unknown
Training can be disabled
DPA available
Xiaomi MiMoMiMo Token Plan Lite
Coding
$6.00or $5.28/month billed annually
Xiaomi MiMo Token Plan Lite: 4.1B credits per month for MiMo-V2.6 Pro and Flash, about 200 tasks per Xiaomi.
Storage
EU (personal data; transfer to Singapore possible)
Model processing
Regional option
No training by default
DPA unknown
StepFunStep Plan Flash Mini
Coding
$6.99or $5.83/month billed annually
StepFun Step Plan Flash Mini: 400M credits per month for Step 5 Preview and Step 3.7 Flash, no 5-hour or weekly windows.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA available
MetaMeta One Core
Image
$7.99
Consumer bundle with more Meta AI image and video generation plus Instagram Plus, WhatsApp Plus and Facebook Plus.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA unknown
CodebuffFreebuff Starter
Coding
Plans are in beta according to the pricing page$8.00or $6.25/month billed annually
Freebuff Starter costs $8 monthly or $75 billed upfront yearly ($6.25/month equivalent). The plans page lists 50 daily hours and a 60-hour monthly bonus.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA unknown
OpenAIChatGPT Go
Chat
$8.00
Low-cost ChatGPT plan on GPT-5.6 Luna with more uploads, images and voice than Free. May include ads.
Storage
US and other countries
Model processing
Unknown
Training depends on conditions
DPA unknown
xAIThe X Subscription Plus
Coding
$8.00or $6.67/month billed annually
Shared weekly Cursor and Grok allowance with confirmed Grok 4.7, 4.6, 4.5 and Composer 2.5 access. Cloud agents, Bugbot, and third-party models are excluded.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA unknown
KuaishouKling AI Standard
Video
$8.80or $6.60/month billed annually
660 credits a month, no watermark, 1080p and 4K, commercial use; about 22 seconds of native 4K video.
Storage
Singapore
Model processing
Unknown
Training depends on conditions
DPA unknown
KreaKrea Basic
Image
$9.00
5,000 compute units a month, all image models, selected video models, and a commercial license. Generation counts are illustrative examples.
Storage
United States and potentially other countries
Model processing
Unknown
Unknown
DPA unknown
StackBlitzBolt.new Lite
App builder
Enroll by October 14, 2026; existing subscriptions continue.$9.00
USD 9 monthly for Forge. Enrollment closes October 14, 2026; existing subscriptions retain the plan. AI image generation is excluded.
Storage
Unknown
Model processing
Unknown
No training by default
DPA unknown
SyntheticSynthetic Go
Coding
$9.00
Synthetic Go costs $9 per month. The subscription includes the currently documented Always-On models and embeddings for API use. Go has no monthly usage reset marked. Its exact weekly allowance and availability in Germany are unknown.
Storage
Unknown
Model processing
Unknown
Unknown
DPA unknown
AdobeAdobe Firefly Standard
Image
$9.99
$0.0999 per video second
Unlimited standard images plus 2,000 credits a month for up to 20 five-second Firefly videos, partner models and audio.
Storage
Unknown
Model processing
Unknown
No training by default
DPA available
ClineClinePass
Coding
$9.99
ClinePass for $9.99 a month: 2 to 5 times the usage of open coding models like Kimi K3 and GLM-5.3 versus API rates.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA unknown
StepFunStep Plan Flash Plus
Coding
$9.99or $8.00/month billed annually
StepFun Step Plan Flash Plus: 1.6B credits per month for Step 5 Preview and Step 3.7 Flash, no 5-hour or weekly windows.
Storage
Unknown
Model processing
Unknown
Training depends on conditions
DPA available
Alibaba Cloud (Qwen)Token Plan Individual Essential
Coding
Limited-time promotional price, list price $16 per month$10.00
Qwen Token Plan Essential: 25,500 credits per month for Qwen3.8, GLM-5.3, DeepSeek-V4 and image models, 2-3 agents at once.
Storage
Singapore
Model processing
Global inference mode
Training depends on conditions
DPA available
BytePlus (ByteDance)ModelArk Coding Plan Lite
Coding
$10.00
ByteDance ModelArk Coding Plan Lite with Dola-Seed, DeepSeek-V4, and GLM-5.3-Flash. Official pages give conflicting approximate request allowances, so the currently applicable amount remains unresolved. OpenClaw and Hermes Agent are listed as integrations.
Storage
Unknown
Model processing
Multiple regions, primarily Asia Pacific
Training depends on conditions
DPA available
Command CodeCommand Code GOAT
Coding
$10.00
Value factor 10.0 ($100 credit)
Command Code GOAT costs $10 per month plus a processing fee and includes up to $100 in monthly plan credits. Model-specific deals can increase usage on the named models; their maxima are not added to the plan-credit balance.
Storage
Unknown
Model processing
US, EU on demand; open source US/EU/Singapore
Training depends on conditions
DPA unknown
GitHubGitHub Copilot access (Standalone)
Coding
Eligible account-team enterprises only; no self-service.$10.00 per user
Value factor 0.0 ($0 credit)
$10 per user without a GitHub Enterprise Cloud license, with no AI credits. Metered Chat and agent requests are billed from first use; inline and next-edit suggestions remain unmetered. Users with a separately paid GHE Cloud license have no additional Copilot seat charge. Eligible account-team enterprises only; no self-service yet.
Storage
Unknown
Model processing
Unknown
Unknown
DPA unknown
Frequently asked questions about AI subscriptions
Start with the tasks and models you need, then compare plans within your budget. Select model families or individual models and use “Every selected LLM” to find plans covering your whole selection. Check usage limits and privacy information alongside the price. A low monthly fee is less useful if the model you need requires extra credits or is unavailable in that plan.
ECB reference rate dated November 2, 2026. €1 = $1.1216, $1 ≈ €0.8916. For plans priced only in US dollars, estimated German gross prices use this rate plus 19% VAT. The amount charged may differ. ECB rate sheet (PDF)
Open the model family or individual model dropdown and check the models you want. “Every selected LLM” shows plans covering your entire selection. With “At least one selected LLM”, a match for one of your selections is enough.
The “Individual models” dropdown contains the “Search models” field. Enter a model name and check the matching result. Your selected models stay visible below the dropdowns, where you can also remove them.
Choose a region under “Hosting location (storage)”. The “Hosting A to Z” sort orders the results by hosting location. Each plan also shows its hosting information directly.
You can filter by hosting location, training policy, EU processing, and DPA availability. Click a plan row or “Show details”. The shared overview contains all model information as well as “Privacy details and sources”.
I separate storage location from model processing. A server in Frankfurt does not tell you where a prompt travels to get an answer.
The hosting filter sorts by documented storage location. The strict EU-only filter also requires documented EU model processing. “EU processing option available” checks your model selection against documented EU routes. General family evidence covers at least one version. You may need to activate the appropriate configuration. A region including Switzerland or other non-EU countries does not count as an EU-only option. A provider’s headquarters does not establish either location.
“No training by default” includes plans that do not use content for model training by default. Plans that let you disable training appear when you select that policy. Any required consent or additional settings are noted for the plan.
A Data Processing Agreement (DPA) sets terms for a provider processing data on your behalf. The directory shows whether a DPA is documented for a particular plan. Hosting regions, training policies, and contracts still require you to assess your actual use case.
“Unknown” means this directory has no verified evidence for the plan. It does not mean the provider lacks the feature or processes data in a particular region.
Yes. The default view shows “Paid plans”. Choose “Free” in the same dropdown for free offers, or “Team” for team plans. “All audiences” shows plans for every audience that match your other filters.
Select fewer models or relax a filter. If access to part of your selection is enough, switch to “At least one selected LLM”. “Reset filters” returns you to the default view.
Changelog
Updates to the AI subscription directory
November 3, 2026v1.28.0
More app routes and corrected credit rules
- Five Workspace plans have documented Lyria 3.5 and Gemini Omni 1.1 Flash routes. Notebook, Gemini App and Vids keep separate allowances and regional conditions.
- Copilot access (Standalone) costs $10 per user/month with no included AI credits. A separately paid GitHub Enterprise Cloud license removes the seat charge. Metered AI requests cost extra; inline suggestions remain unmetered.
- Manus Chat Mode now consumes credits. Free can use at most 1,500 daily-refresh credits per month; subscription credits renew monthly without rollover.
- Claude Duet is recorded as a US waitlist preview. Astra in Work and Codex draws from limited or full existing usage according to the ChatGPT plan.
November 2, 2026v1.27.0
Workspace plans and clearer agent costs
- Google Workspace distinguishes basic access from higher Gemini App allowances. Fast, Pro and Thinking do not establish exact model versions; image and video routes are recorded separately.
- OpenCode lists DeepSeek V4.1 Flash monthly limits of $100, $200 and $105 for Go, Go Plus and Go Teams. The DeepSeek ZDR agreement runs through November. Conflicting or currently unconfirmed routes retain unknown access.
- Copilot has separate compute, memory and snapshot meters for cloud sandboxes. Dollar-billed Warp accounts incur platform-time and hosted-compute charges according to run conditions.
- OpenAI documents a default 500,000-token context for GPT-6.1 Sol in Work and Codex. The long-context surcharge exception applies only within included subscription usage.
November 1, 2026v1.26.0
Timed models and upcoming allowances
- Ling 3.1 Flash 1019 is free until November 3 at 16:00 UTC, capped at 100 requests/day and requiring a $1 credit balance. Separate Glyph Cluster and Laguna S 2.1 CLI limits are 1,000 requests/day.
- Langdock includes 2,500 workflow runs per workspace/month. Model usage costs extra; Governance has a separate temporary free period followed by per-seat pricing.
- OpenAI announces a shared ChatGPT/Codex token allowance for Plus, Pro, and Business in December. Instant chat is excluded; current allowances remain until the transition.
October 31, 2026v1.25.0
Claude Ultra and clearer app costs
- Claude Ultra replaces Max 20x at the same price. 10x Pro weekly usage, with no five-hour limit for verified accounts. Separately claimed API credits do not extend Chat usage.
- Krea Ad Studio uses Krea 3 for static ads and consumes compute units. Image generation requires a paid plan from Basic upward.
- GitHub says Copilot cloud sandboxes will use the separate Actions compute allowance from November 2026. Estimated euro prices use the October 30, 2026 ECB rate.
October 30, 2026v1.24.0
ChatGPT Work and Codex allowances
- Pro 100 and Pro 500 offer twice their previous Work/Codex allowances. Pro 200 has one allowance across subscribers; the announced reduction is canceled.
- The outdated five-hour Pro estimate is removed. Business Premium’s announced increase is not converted into a new Standard-relative multiplier.
October 30, 2026v1.23.0
Student plan and clearer usage conditions
- Kiro Students with one free year and 1,000 monthly credits for eligible students. Model and privacy limits remain explicit.
- Announced Qwen 3.5 Plus retirement, confirmed Haiku 4.5 removal from Claude chat, and more precise Otter workflow eligibility.
- Seedance 2.5 costs on Krea and Magnific, with separately billed Draft and final passes. Conflicting Krea image examples are excluded from image-price calculations.
- Sourcegraph credit rollover and conditional Kilo offers. Estimated euro prices use the ECB rate from October 29, 2026.