More AI assistant and coding agent plan limits

Credits replaced by quotas in March 2026

In March 2026 the credit-based system was replaced with a daily and weekly usage allowance that refreshes automatically.

Quota resets on calendar dates

The quota resets daily and weekly, based on the calendar date.

Free tier model restriction

Free tier users only have access to SWE-1.6.

SWE-2 free only until November 30, 2026

SWE-2 is free for self-serve plans in Devin Desktop and Devin CLI through November 30, 2026, not in Devin cloud sessions; list pricing applies afterward.

Fast modes cost more

Priority and speed configurations (e.g. SWE-1.5 Fast, fast Opus variants) increase the cost.

On-demand credits roll over

Purchased on-demand credits roll over month-to-month and never expire.

No free trials of paid plans

Free trials of paid plans are not offered; a refund can be requested before using the subscription.

SSO not in new Teams plans

New Teams plans do not include SSO; it is an Enterprise-only feature.

Flex seats exclude Devin Desktop

Teams flex seats draw from shared on-demand credits and do not include Devin Desktop access.

Teams Automations and Review use shared credits

On Teams, Automations and Devin Review draw from the shared on-demand credit pool, not full-seat quota; if credits run out, Automations stop.

Pro and Max are single-user

Pro and Max are individual plans and cannot be shared across multiple users.

Failed AI requests

Not captured

Refunds

Not captured

Autocomplete limited on lower enterprise tier

One enterprise tier lists limited autocomplete and IDE / Plugin access; the other lists unlimited autocomplete and adds CLI.

Spend limit hard-stops Pro usage

Once the Pro monthly spend limit is hit, further usage stops until the token spend limit resets.

Business spend limit starts at $0

The org-wide AI spend limit starts at $0 and must be increased before members can use any hosted models.

Trial limited to one model

GPT-6 Luna is the only hosted model available during the trial. The trial ends when the $5 balance is used or 14 days pass, whichever comes first. The $5 trial balance is shared across Zed and Delta.

Student plan model restrictions and no spend limit

Student plan excludes Claude Fable, Claude Opus, GPT-6 Astra, GPT-5.5 pro, and GPT-5.4 pro, and cannot configure spend limits; usage is capped at the $10/month included credit.

Monthly credit resets on billing date

Monthly included credit resets on your monthly billing date. No rollover is mentioned.

Business seats include no AI credit

Business seats don't bundle a fixed LLM credit allotment; Zed-hosted AI is billed at standard usage rates, or teams can bring their own API keys.

No Business free trial; SSO not available

Business doesn't currently have a free trial. SSO, SAML, and SCIM are planned but not currently available.

Gemini 3.1 Pro context capped

Zed currently limits hosted Gemini 3.1 Pro requests to 200k tokens because pricing changes above that context size.

Payment failure blocks hosted models

If payment of an invoice fails, Zed blocks usage of hosted models until the payment is complete.

Unused quota does not roll over

Unused included quota does not roll over and resets every 30 days. Purchased top-up credits stay valid for 12 months.

Reset date may not match billing date

The included quota resets every 30 days from the date you first activated your JetBrains AI license, which may not match your billing cycle.

Code completion does not use credits

Code completion and next edit suggestions use JetBrains' own models and do not use AI quota.

Usage varies with tokens, model and agents

Each interaction costs a different amount, based mainly on tokens. Long chat threads, high-cost models (marked with an icon) and heavy use of agents use up quota faster.

Top-up credits used only after monthly quota

Purchased credits are used only after the monthly quota is fully used. When the quota renews, it is used first again.

AI Free not available in some IDEs

AI Free is not available in Android Studio, IntelliJ IDEA without an Ultimate subscription, or PyCharm without a Pro subscription.

AI Pro included with some subscriptions

AI Pro is included at no extra cost with All Products Pack and dotUltimate (not the free versions for students, teachers, classrooms or Open Source programs), with Developer Recognition and Content Creators programs, and with DataSpell (DataSpell only).

Version and region limits

AI Free needs IDE version 2025.1 or later. AI Ultimate needs 2024.2.1 or later. The JetBrains AI service is limited to allowed territories and has not been available in China since July 13, 2026.

Organizations moving to a shared credit pool

In organizations moved to JetBrains Central, credits go into one pool shared across the organization. Admins set per-user monthly limits, and usage is counted per calendar month instead of a rolling 30-day cycle.

Air seats do not cover usage

On the Air Business and Enterprise plans, an AI seat is needed for each user with AI access. The seat does not cover usage, which is charged separately in AI credits.

Included usage does not roll over

Included usage resets monthly, even on annual subscriptions, and unused included usage does not roll over.

Purchased usage expires after 12 months

Purchased usage rolls over across billing cycles and expires 12 months after purchase.

Existing seats may have a different allowance

Existing seats can have a different allowance from new subscriptions. Use the amount shown for your account.

Spend limit is not a guaranteed ceiling

The spend limit is not a guaranteed ceiling on the amount charged. Free-plan premiums and taxes can increase the payment total.

Auto-reload can repeat under heavy use

Heavy usage by any member can trigger repeated reloads.

Cloud runs add platform and compute charges

Cloud runs incur platform charges for billable agent time, plus compute charges on Warp-hosted compute, in addition to inference.

BYO inference on Business/Enterprise local runs still incurs platform charges

Local runs on Business and Enterprise incur platform charges when they use customer-supplied inference. Local runs on Free, Build, or Max have no platform charge.

BYOK limited by company size

BYOK and custom inference endpoints are available for individuals and organizations with 10 or fewer employees. Larger organizations need a Business or Enterprise plan.

Usage per prompt varies

Cost depends on model, tokens, caching, task complexity and provider tools. Two similar prompts can use different amounts.

One plan per team

All team members have the same subscription plan; plans cannot be mixed within a team.

Advanced model access can be limited in heavy weeks

For Advanced AI model queries, access may be limited during weeks of especially heavy usage.

Model restrictions by plan

Free has no access to advanced AI models. o3-Pro and Claude 4.5 Opus thinking models are listed for Max users.

Monthly credits do not roll over

Unused monthly credits expire at the end of each billing cycle; annual subscribers still refresh monthly.

Purchased credits expire after a year of inactivity

Purchased credits expire if you don't use Computer for one year, and can only be used with an active paid plan.

Pro weekly Computer usage: Light effort only without credits, no rollover

With included usage but no credits, Computer runs on the Light effort setting. Included usage resets weekly from first use and does not roll over.

Weekly included usage not on promo, Max or Enterprise plans

Included weekly usage is only on paid Consumer Pro plans.

Comet Work mode consumes credits

Ask searches and Comet Assistant in Search mode do not use credits; Computer tasks and Comet Assistant in Work mode do.

Negative credit balance possible

The step already running when credits run out is still billed, so the balance can go below zero.

Max annual billing only on web; mobile upgrades create separate subscription

Annual billing for Max is only available on the web app. Upgrading via mobile app creates a new separate Max subscription in addition to an ongoing web Pro subscription.

Plan model access does not cover API

Advanced model access applies within Perplexity web only; API access is billed separately.

Credit bonus is one per account

The limited-time sign-up bonus does not repeat on resubscribe, billing switch, or discount.

Trials require a card and auto-renew

A valid card is required for promotional and trial redemptions; the subscription auto-renews at the standard rate unless cancelled.

Refund windows

Outside EU/UK/Turkey (14 days) and South Korea/Brazil (7 days), refund requests are accepted within 24 hours for monthly and 48 hours for annual Max subscriptions.

Organization File Repository retired

The Organization File Repository is being retired on October 1, 2026; remaining files will be deleted.

Fair usage limits on paid plans

Pro and Team prices are subject to fair usage limits and Mistral's Terms of Service.

Le Chat renamed to Vibe

Le Chat has become Vibe. Account, plan, conversations and saved content carry over. Same URL and login (chat.mistral.ai).

Education plan eligibility

Valid for up to 12 months for verified students at accredited higher-education institutions who have not used Vibe (or Le Chat) before. Requires a verified institutional email address; student IDs and enrollment documents are not accepted.

Annual billing discount

Pro and Team can be billed monthly or annually; annual billing includes a 20% discount compared with monthly billing.

API pay-as-you-go is separate from the subscription

Pay-as-you-go for Mistral APIs is a separate setting from your subscription.

Refunds only within 14 days of initial purchase

Web subscriptions are refundable only if cancelled within 14 days of the initial purchase; this does not apply to renewals. App Store and Google Play refunds are handled by the store.

Model training opt-out

Model training is opt-out on Pro, Team and Enterprise.

Monthly credits do not roll over

Unused monthly credits are cleared at the start of each cycle, on both monthly and annual subscriptions.

Daily credits expire every 24 hours

The daily balance resets to 300 at 00:00 UTC, and unused daily credits are removed.

Daily credits limited to Manus 2.0 lite

Daily refresh credits can only be used for Manus 2.0 lite.

Credit consumption order

Credits are used in this order: event, daily, monthly, add-on, then free credits.

Add-on credits need an active paid subscription

Add-on credits carry over while a paid subscription is active and become inactive if the account switches to Free or the subscription lapses.

Balance can go negative

Each action's cost is deducted at the start of the next action, so the final action of a task can take the balance below zero. The negative balance is resolved at the next refresh.

Credits lost on cancellation

Membership benefits, including access to credits, end as soon as the subscription ends. Monthly credits that were reset to 0 cannot be reinstated.

Free trial cancellation

With the 7-day free trial, cancelling before the charge costs nothing. Cancelling after the charge gives no refund and only cancels the next renewal.

Mid-cycle upgrades

Upgrading does not change the cycle end date. For the remaining time in the current cycle, the difference in credits between the two plans is issued.

Team credits are pooled

Credits are shared across all seats in the team.

Legacy plans renamed

Existing members keep their prices on auto-renewal. The original Basic and Plus plans were renamed to Pro.

Cue phone number billed separately

A Cue phone-number subscription is purchased separately. Phone usage counts toward a Cue weekly allowance, not Manus credits.

Message counts are estimates, not fixed limits

The 5-hour figures are estimates for local messages; cloud tasks may use more of the allowance; usage depends on model and task.

Shared allowance with ChatGPT Work and other agentic features

ChatGPT Work and Codex share usage; limits are also shared with other agentic features such as ChatGPT for Excel on Plus and Pro.

Fair use on in-progress turns

If you hit your limit during an active turn, the agent can continue that turn, subject to fair use limits.

Ultrafast only on Pro $500 and eligible Enterprise/Edu

Other self-serve plans don't have Ultrafast at launch, even with purchased credits.

Model restrictions on Free and Go

Free and Go list GPT-6 Luna at Standard speed in the desktop app, subject to rollout.

GPT-5.5 retirement

GPT-5.5 retires from Codex with ChatGPT sign-in on October 14, 2026; API-key sessions are not affected.

API key mode has no cloud features

API-key Codex has no cloud-based features and is billed at API pricing.

Code Review via GitHub counted separately

Code Review usage applies only when Codex runs reviews through GitHub; reviews run locally or outside GitHub count toward general usage limits.

Promotional GPT-5.6 Sol pricing

GPT-5.6 Sol promotional pricing is available at least through November 21, 2026.

ClinePass covers only curated open-weight models

ClinePass includes a curated list of open models (GLM, Kimi, DeepSeek, MiMo, MiniMax, Muse Spark, Qwen). Other models go through Cline (usage-billing) at pay-as-you-go API rates, or through your own API keys.

Three usage windows on ClinePass

Usage is measured against a rolling 5-hour window, the calendar week and the calendar month.

Models removed from ClinePass

Due to capacity constraints and model upgrades, GLM-5.2, Kimi K2.6, Kimi K2.7 Code and DeepSeek V4 Flash are no longer available on ClinePass.

Promotional pricing and processing fees

Standard rate is $9.99 per month after the promotion period; additional processing fees may apply.

DeepSeek V4 Pro peak and off-peak rates

DeepSeek V4 Pro has separate peak and off-peak reference rates.

Free models not available via API

Free model usage works only in the Cline IDE Extension and CLI, not through the Cline API. Free model usage may be used to help improve model performance and quality.

ClinePass usable in other agents

ClinePass can be used with other coding agents through a ClinePass API key.

Cancellation

Cancel in account settings before the next renewal; access continues until the end of the current term.

Credits may expire after one year

OpenRouter reserves the right to expire unused credits one year after purchase.

24-hour refund window; fees non-refundable

Unused credits can be refunded only within 24 hours of purchase. Platform fees are not refunded, and crypto payments are never refundable.

BYOK can fall back to paid shared capacity

By default, if all your BYOK keys hit a rate limit or fail, requests fall back to shared OpenRouter endpoints, which spend OpenRouter credits.

BYOK allowance is measured by cost, not requests

The fee-free BYOK allowance is measured by list-price inference cost, not by request count.

In-flight spending budget

Concurrent paid requests are limited by an in-flight budget: a fraction of your credit balance, up to a fixed ceiling. Requests that do not fit are rejected even if your balance is positive.

Negative balance blocks free models

With a negative credit balance you may see errors, including on free models.

Extra accounts or keys do not raise limits

Rate limits are governed globally; extra accounts or API keys do not change them.

Features gated by plan

Free has no auto-routing, prompt caching or budgets. EU/US in-region routing and advanced notifications need Business or Enterprise. SSO/SAML and SLAs are Enterprise only.

Credits are pooled across the organization or enterprise

Each user's included AI credits go into one pool at the billing entity level. For example, 100 Copilot Business users share a pool of 190,000 AI credits.

No rollover; fixed monthly reset

Included AI credits do not carry over between months. Unused credits are forfeited, and the pool resets at 00:00:00 UTC on the first day of each calendar month.

Removing seats mid-cycle doesn't shrink the pool until next cycle

Adding licenses mid-cycle increases the pool immediately. Removing licenses mid-cycle takes effect at the start of the next billing cycle.

Paid overage is on by default

To prevent any spending beyond included AI credits, an administrator must explicitly disable the AI credits paid usage policy in AI Controls settings.

Additional usage may be capped

If the additional-usage cap is reached, the additional usage already consumed must be paid off before usage can continue.

User-level budgets are hard stops

User-level budgets cap an individual's total consumption per billing cycle, including usage from the included pool. They always enforce a hard stop. A $0 user-level budget immediately blocks features that consume AI credits.

What consumes credits

Copilot Chat, Copilot CLI, Copilot cloud agent, Copilot Spaces, Spark and third-party coding agents consume AI credits. Cost depends on the model and the number of tokens used.

Completions keep working when blocked

Inline completions and next edit suggestions remain available when a user has a $0 budget or reaches an enforced spending limit.

Model access differs by plan

Copilot Business has access to premium models. Copilot Enterprise and Copilot access have priority access to premium models.

One billing model per enterprise

Copilot access cannot coexist with Copilot Business or Copilot Enterprise seats in the same enterprise. Organizations within the enterprise cannot choose a different billing model.

Not available on GitHub Enterprise Server

Copilot is not currently available for GitHub Enterprise Server.

Claude Fable models retain data by default

When Claude Fable 5 or 5.1 is used, Anthropic retains prompts and outputs by default for safety classifiers. Zero data retention is available on request through the end of 2026. An administrator must enable each model.