DevTk.AI

AI Coding Tool Plans, Limits and Overage

What does a $20 or $200 coding plan actually include, and what happens when you hit the limit? This page compares the official plans for Claude Code, OpenAI Codex, Cursor, GitHub Copilot, Qoder and Devin. The bracketed numbers in each cell point to the official source at the bottom of the page, verified 2026-11-16. Anything a vendor does not publish is marked "not published" rather than guessed.

Claude Code

Anthropic

Claude Code is included in Claude Pro, Max, Ultra and Team. Limits are shared with Claude chat: a 5-hour session limit plus a weekly limit (verified Ultra accounts have no 5-hour limit).

OpenAI Codex

OpenAI

Codex is included in every ChatGPT plan and shares one allowance with ChatGPT Work. OpenAI publishes estimated local messages per 5 hours for Plus; Pro plans currently have no 5-hour limit, and weekly limits may apply.

Cursor

Cursor

Paid Cursor plans include one weekly pool for Cursor's own models (Grok 4.7, 4.6, 4.5 and Composer 2.5), sized as a multiple of Plus. Third-party models such as Claude, GPT and Gemini are not in the pool.

GitHub Copilot

GitHub

Since 2026-06-01 Copilot bills chat, agent, CLI and code review in GitHub AI Credits (1 credit = $0.01), consumed at each model's API rates. Code completions and next edit suggestions do not use credits.

Qoder

Qoder

Qoder plans include a monthly Credits allowance. Plan Credits expire at the end of the billing period and do not roll over.

Devin (Windsurf)

Cognition

windsurf.com/pricing now redirects to devin.ai/pricing. Paid plans refresh their allowance daily and weekly; Cognition does not publish the quota sizes.

Plan comparison

Prices are USD per month on monthly billing (annual or per-seat pricing in the note). The cited official pages show no CNY price for these tiers, so prices are USD only. Bracketed numbers link to the official sources at the bottom.

Plan Price Models Allowance Reset window At the limit Own API key / third-party models
Claude Code · Anthropic Official pricing ↗
Pro $20/mo [1]
$17/mo billed annually ($200 up front)
Opus, Sonnet and Haiku; Fable is paid with usage credits[1] 1x Pro usage (Anthropic's baseline)[1] 5-hour session limit + weekly limit[1] Wait for the reset, upgrade, or turn on usage credits (billed at standard API rates). A response already in progress can wrap up once per week.[2] An API key or gateway credential replaces the subscription: that traffic is billed per token to the key owner and plan limits don't apply. Anthropic does not support non-Claude models in Claude Code.[4]
Max $100/mo [1] Opus, Sonnet and Haiku; Fable up to half the weekly limit[1] 5x Pro usage; $100/mo in API credits for the Claude Platform[1] 5-hour session limit + weekly limit[1] Wait for the reset, upgrade, or turn on usage credits (billed at standard API rates).[3] Same as Pro: an API key bypasses plan limits and bills per token.[4]
Ultra $200/mo [1] Opus, Sonnet and Haiku; Fable up to half the weekly limit[1] 10x Pro usage per week; $200/mo in API credits for the Claude Platform[1] Weekly limit only (no 5-hour limit for verified accounts)[1] Wait for the weekly reset or turn on usage credits (billed at standard API rates).[2] Same as Pro: an API key bypasses plan limits and bills per token.[4]
Team (Standard seat) $25/mo [1]
Per seat; $20/seat billed annually; 2-150 seats
All Claude features including Claude Code[1] "More usage than Pro"; up to $500/mo in pooled API credits per team[1] Not published by Anthropic[1] Usage credits are available on Team.[1] Same as Pro: an API key bypasses plan limits and bills per token.[4]
Team (Premium seat) $125/mo [1]
Per seat; $100/seat billed annually
All Claude features; Fable up to 50% of weekly limits[1] 5x the usage of a Standard seat[1] Not published by Anthropic[1] Usage credits are available on Team.[1] Same as Pro: an API key bypasses plan limits and bills per token.[4]
OpenAI Codex · OpenAI Official pricing ↗
ChatGPT Free $0/mo [5] GPT-6 Luna at Standard speed in the desktop app (subject to rollout)[5] Not published by OpenAI[5] Not published by OpenAI[5] Run extra local chats with an API key at standard API rates.[5] API-key sign-in works in the CLI, SDK and IDE extension, billed at API pricing.[5]
ChatGPT Go $8/mo [5] GPT-6 Luna at Standard speed in the desktop app (subject to rollout)[5] Not published by OpenAI[5] Not published by OpenAI[5] Run extra local chats with an API key at standard API rates.[5] API-key sign-in works in the CLI, SDK and IDE extension, billed at API pricing.[5]
ChatGPT Plus $20/mo [5] GPT-6.1 Sol and GPT-6 Luna; Codex web, CLI, IDE extension, iOS[5] Estimated local messages per 5 hours: GPT-6.1 Sol 15-160, GPT-6.1 Terra 35-285, GPT-6 Luna 350-3,000. Cloud tasks use more.[5] 5-hour window; weekly limits may also apply[5] The active turn can finish (fair use). Buy ChatGPT credits, or use an API key at standard API rates.[5] API key billed at API pricing; Codex config also supports custom model providers (e.g. DeepSeek's documented setup), billed by that provider.[6]
ChatGPT Pro $100/mo [5]
Three tiers: $100, $200 or $500 per month; Ultrafast mode on Pro $500
Everything in Plus[5] Higher than Plus; per-tier allowances are not published on the Codex pricing page[5] No 5-hour limit; weekly limits may apply[5] The active turn can finish (fair use). Buy ChatGPT credits, or use an API key at standard API rates.[5] Same as Plus.[6]
ChatGPT Business $25/mo [5]
Per user; $20/user billed annually; 2+ users
Codex across desktop and mobile; larger cloud VMs[5] Standard Business: same per-5-hour message estimates as Plus[5] 5-hour window; weekly limits may also apply[5] Workspaces with flexible pricing can buy workspace credits; anyone can use an API key at standard API rates.[5] Same as Plus.[6]
Cursor · Cursor Official pricing ↗
Free $0/mo [7] Agent and Chat[7] "Limited usage"; size not published[7] Not published by Cursor[7] Upgrade to a paid plan.[7] Not published by Cursor[7]
Plus $8/mo [7]
$80/year
Cursor models only (Grok 4.7, 4.6, 4.5, Composer 2.5) at fixed settings: Fast off, medium effort[7] Base weekly pool (1x)[7] Weekly[7] Wait for the weekly reset or upgrade.[7] No third-party models and no own API keys on Plus.[7]
Premium $30/mo [7]
$300/year
Cursor models in the pool; Claude, GPT, Gemini and other third-party models outside it[7] 4x the Plus weekly pool[7] Weekly[7] Wait, use prepaid credit (optional Auto Top-up, monthly additional-usage cap), or upgrade. Requests are not downgraded.[7] Third-party models bill at API list price from the first request, from prepaid credit. Individual plans don't pay the Cursor Token Rate.[7]
Super $100/mo [7]
$1,000/year
As Premium, plus priority model access at peak times[7] 15x the Plus weekly pool[7] Weekly[7] Wait, use prepaid credit, or upgrade.[7] Same as Premium.[7]
Ultra $200/mo [7]
$2,000/year
As Super, plus earliest access to new products and models[7] 50x the Plus weekly pool[7] Weekly[7] Wait or use prepaid credit.[7] Same as Premium.[7]
Teams $30/mo [7]
$30, $100 or $200 per seat/month (seat tier matches Premium, Super or Ultra)
As the matching individual tier[7] Matches each seat's tier[7] Weekly[7] Admins manage a shared prepaid balance and one spending cap.[7] Eligible third-party and BYOK requests carry a Cursor Token Rate of $0.25 per 1M tokens.[7]
GitHub Copilot · GitHub Official pricing ↗
Free $0/mo [8] Entry-level models[8] 2,000 completions/month plus a limited AI Credits allowance (amount not published)[8] Monthly[8] Wait for the next cycle; buying extra credits is not available on Free.[8] User-configured BYOK in VS Code, JetBrains, Xcode, Copilot CLI and the app sends requests directly to your provider.[10]
Pro $10/mo [9] Model selection; unlimited completions and next edit suggestions[9] 1,500 AI Credits/month (1,000 base + 500 flex) = $15 of usage[9] Monthly[9] Set a dollar budget for additional usage, wait for the reset, or switch to a cheaper model.[8] User-configured BYOK; requests go directly to your provider and are billed by it.[10]
Pro+ $39/mo [9] Premium models including Opus[9] 7,000 AI Credits/month (3,900 base + 3,100 flex) = $70 of usage[9] Monthly[9] Set a dollar budget for additional usage, wait for the reset, or switch to a cheaper model.[8] Same as Pro.[10]
Max $100/mo [9] Priority access to new models and features[9] 20,000 AI Credits/month (10,000 base + 10,000 flex) = $200 of usage[9] Monthly[9] Set a dollar budget for additional usage, wait for the reset, or switch to a cheaper model.[8] Same as Pro.[10]
Business $19/mo [9]
Per granted seat
Subject to administrator policies[9] 1,900 AI Credits per user per month[9] Monthly[9] Admins decide whether additional paid usage is allowed; if not, Copilot pauses until the next cycle.[8] Enterprise/org custom models are in public preview; admins can disable user-configured BYOK.[10]
Enterprise $39/mo [9]
Per granted seat
Subject to administrator policies[9] 3,900 AI Credits per user per month[9] Monthly[9] Admins decide whether additional paid usage is allowed; if not, Copilot pauses until the next cycle.[8] Same as Business.[10]
Qoder · Qoder Official pricing ↗
Free $0/mo [11] Basic models; 2-week Pro trial with 300 Credits for new users[11] Limited user messages, with daily and monthly limits[11] Daily and monthly[11] Upgrade to a paid plan.[11] BYOK is listed on Free.[11]
Lite $10/mo [11] Curated models; full agent capabilities[11] 1,000 Credits/month[11] Monthly billing period, no rollover[11] Upgrade, or buy Credit Packs ($20 per 1,500 Credits, valid one month).[11] Custom models from Alibaba Cloud Model Studio, DeepSeek, Z.ai, Kimi, MiniMax and Xiaomi MiMo bill to your provider account, not Qoder Credits (Repo Wiki excepted).[12]
Pro $20/mo [11] Full model lineup; Repo Wiki and Experts Mode[11] 2,000 Credits/month[11] Monthly billing period, no rollover[11] Upgrade, or buy Credit Packs ($20 per 1,500 Credits, valid one month).[11] Same as Lite.[12]
Pro+ $60/mo [11] Full model lineup[11] 6,000 Credits/month[11] Monthly billing period, no rollover[11] Upgrade, or buy Credit Packs ($20 per 1,500 Credits, valid one month).[11] Same as Lite.[12]
Ultra $200/mo [11] Full model lineup[11] 20,000 Credits/month[11] Monthly billing period, no rollover[11] Buy Credit Packs only ($20 per 1,500 Credits, valid one month).[11] Same as Lite.[12]
Devin (Windsurf) · Cognition Official pricing ↗
Free $0/mo [13] Not published by Cognition[13] "Light quota to code with agents"[13] Not published by Cognition[13] Not published by Cognition[13] Not published by Cognition[13]
Pro $20/mo [13] Includes Devin Cloud[13] Higher quotas than Free; size not published[13] Refreshes daily and weekly[13] Purchase extra usage at API pricing.[13] Not published by Cognition[13]
Max $200/mo [13] Includes Devin Cloud[13] "Significantly higher quotas" than Pro; size not published[13] Refreshes daily and weekly[13] Purchase extra usage at API pricing.[13] Not published by Cognition[13]
Teams $80/mo [13]
$80/month for the team plus $40/month per full dev seat; up to 100 users
Not published by Cognition[13] Cost controls with 3 usage tiers; size not published[13] Refreshes daily and weekly[13] Purchase extra usage at API pricing.[13] Not published by Cognition[13]

What happens when you hit the limit

"Out of usage" means different things in each tool: some stop, some keep going and bill at API rates, some sell separate credits. Here is what each vendor documents.

Claude Code

Optional usage credits keep you working after a limit, billed at standard API rates, with a monthly spend cap you set.

Wait for resetUpgradePay-as-you-go overage

OpenAI Codex

Plus and Pro can buy ChatGPT credits; anyone can switch to an API key billed at standard API rates.

Switch to API keyUpgradeBuy credits

Cursor

On Premium and above, third-party models bill at the model's API list price from the first request, paid from prepaid credit; Cursor models past the pool bill at Cursor's published per-token rates.

UpgradeWait for resetPay-as-you-go overage

GitHub Copilot

Set a dollar budget for additional usage ($10 covers 1,000 credits), wait for the monthly reset, or switch to a cheaper model.

Wait for resetUpgradePay-as-you-go overagePauses

Qoder

Upgrade, or buy Credit Packs at $20 per 1,500 Credits (each pack expires one month after purchase). Ultra can only buy packs.

Wait for resetUpgradeBuy credits

Devin (Windsurf)

Paid plans can purchase extra usage at API pricing.

UpgradeWait for resetPay-as-you-go overage

Things that catch people out

  • Claude Code shares its limits with Claude chat. Once an API key or gateway credential is set, Claude Code stops using the subscription and bills per token.
  • Codex usage on an API key is separate from subscription usage, and OpenAI says not to use API token prices to estimate included tasks.
  • On Cursor's new plans, Claude, GPT, Gemini and other third-party models bill at API list price from the first request, from prepaid credit, not from the weekly pool.
  • GitHub Copilot AI Credits are consumed at each model's API rates, so expensive models drain the allowance faster. Code completions don't use credits.
  • Qoder plan Credits and Credit Packs both expire and do not roll over.

2026 plan and billing changes

Only changes we could verify on an official vendor page. Where the vendor gives no date, we show the date we first recorded it.

Date Tool Change
2026-11-16
first recorded
Cursor Cursor's pricing docs list new plans (Free, Plus $8, Premium $30, Super $100, Ultra $200) with a weekly pool for Cursor models and third-party models billed at API list price. Hobby, Start, Pro, Pro+, the old Ultra and Teams Standard/Premium seats continue for existing customers. Cursor does not publish the switch date. Cursor docs: models & pricing ↗
2026-10-16
effective
GLM Coding Plan (Z.AI) Standard API keys and GLM Coding Plan keys are now managed separately. For keys created from this date, the key type decides billing: a standard key used in a coding tool bills pay-as-you-go and does not use Coding Plan quota. Z.AI API key management guide ↗
2026-10-14
effective
OpenAI Codex GPT-5.5 retired from Codex with ChatGPT sign-in at 19:00 UTC. API-key sessions were not affected. Codex pricing ↗
2026-10-07
announced
Claude Code Max, Ultra and Team plans now include monthly API credits for the API, Managed Agents, and the Agent SDK / `claude -p` when used with an API key from a linked Console organization. Claude Help Center ↗
2026-06-15
announced
Claude Code Anthropic paused previously announced changes to Agent SDK usage: the Agent SDK, `claude -p` and third-party apps still draw from subscription limits. Claude Help Center ↗
2026-06-01
effective
GitHub Copilot Usage-based billing took effect. Annual Pro and Pro+ subscribers stay on premium-request pricing until their plan expires; code completions and next edit suggestions remain included. GitHub changelog ↗
2026-04-30
effective
Qoder The 50% individual promotion ended: Pro $10 → $20, Pro+ $30 → $60, Ultra $100 → $200 per month (Credits unchanged). Teams seats went from $30 to $40 with 3,000 Credits instead of 2,000. Qoder pricing adjustment notice ↗
2026-04-27
announced
GitHub Copilot GitHub announced that all Copilot plans move to usage-based billing on 2026-06-01: premium request units are replaced by GitHub AI Credits, consumed at each model's published API rates (input, output and cached tokens). GitHub changelog ↗

Model API price change log →

Subscription vs API: the same usage at API list prices

Most of a coding agent's input is repository context sent again and again, so the cache-hit rate drives the bill. This table is an estimate with these assumptions:

  • 90% of monthly input tokens are cache hits billed at the cached-input rate; the other 10% bill at the standard input rate.
  • Anthropic and OpenAI cache-write premiums, long-context surcharges, Fast mode and web-search fees are not included.
  • DeepSeek uses its peak rate (weekdays 01:00-04:00 and 06:00-10:00 UTC); off-peak is 50% lower.
  • Prices come from DevTk.AI's model price data, read at build time.

Subscription plans do not publish token allowances, so this table cannot tell you what a plan is "worth" in API dollars. It shows what your volume would cost if you paid the API directly.

Model Input / cached / output (per 1M) Light
20M in + 0.4M out / mo
Daily
100M in + 2M out / mo
Heavy
400M in + 8M out / mo
Claude Opus 5.5
Claude Code default model
price verified 2026-11-12
$4.00 / $0.2 / $20.00 $19.60$98.00$392
Claude Sonnet 5.5
price verified 2026-11-12
$2.00 / $0.1 / $10.00 $9.80$49.00$196
GPT-6.1 Sol
price verified 2026-11-16
$2.00 / $0.1 / $10.00 $9.80$49.00$196
GPT-5.3-Codex
price verified 2026-11-16
$1.75 / $0.175 / $14.00 $12.25$61.25$245
GPT-6 Luna
price verified 2026-11-16
$0.1 / $0.01 / $0.5 $0.58$2.90$11.60
Kimi K3
price verified 2026-11-16
$3.00 / $0.3 / $15.00 $17.40$87.00$348
GLM-5.3
price verified 2026-11-16
$1.40 / $0.26 / $4.40 $9.24$46.20$185
DeepSeek V4 Pro (Peak)
price verified 2026-11-16
$1.32 / $0.044 / $3.96 $5.02$25.08$100
DeepSeek V4.1 Flash (Peak)
price verified 2026-11-16
$0.3 / $0.006 / $1.20 $1.19$5.94$23.76
GLM-5.3 Flash
price verified 2026-11-16
$0.15 / $0.03 / $0.5 $1.04$5.20$20.80

Formula: monthly cost = input × 10% × input price + input × 90% × cached price + output × output price. To plug in your own numbers, use the pricing calculator.

Run your coding tool on a cheaper model

These guides cover only setups that the model vendor or the tool documents officially (official Anthropic-compatible or OpenAI-compatible endpoints). No relays, resellers or workarounds.

AI coding agent API cost comparison → · Coding tool deals and free credits →

Official sources

  1. [1] Claude Code: Claude pricing · verified 2026-11-16
  2. [2] Claude Code: Use Claude Code with your Pro, Max, or Ultra plan · verified 2026-11-16
  3. [3] Claude Code: Manage usage credits for paid Claude plans · verified 2026-11-16
  4. [4] Claude Code: Claude Code docs: LLM gateways · verified 2026-11-16
  5. [5] OpenAI Codex: Codex pricing · verified 2026-11-16
  6. [6] OpenAI Codex: Codex docs: advanced config (model providers) · verified 2026-11-16
  7. [7] Cursor: Cursor docs: models & pricing · verified 2026-11-16
  8. [8] GitHub Copilot: GitHub Copilot plans · verified 2026-11-16
  9. [9] GitHub Copilot: GitHub Docs: Copilot plans · verified 2026-11-16
  10. [10] GitHub Copilot: GitHub Docs: bring your own key · verified 2026-11-16
  11. [11] Qoder: Qoder docs: pricing · verified 2026-11-16
  12. [12] Qoder: Qoder docs: custom models · verified 2026-11-16
  13. [13] Devin (Windsurf): Devin pricing · verified 2026-11-16

Spotted something out of date? Tell us on the feedback page and we will re-check it against the official page. Feedback