Plans

Price, official limits, and what builders actually report. Limits move fast — check dates and vendor pages before buying.

Prices and official limits are checked against each vendor's own page (linked on every card). Value scores are AgentMPG editorial estimates, not measurements.

Anthropic · Claude Code

Claude Max (was Max 5x)

$100/mo
Featured

Est. value score (editorial estimate)

88
Official limits
Renamed from Max 5x in Nov 2026 (same price). 5x Pro usage every rolling 5-hour session and every week. Usage is shared across Claude chat and Claude Code. Fable (Anthropic's top model) included for up to half your weekly limit. Includes $100/month in Claude Platform API credits and priority access at peak times. Billed monthly only. Claude Duet (US research-preview waitlist on Pro and Max) shares these limits, and its background check-ins count toward them.
Source
claude.com/pricing · verified 2026-11-02
Crowd-reported burn
Sweet spot for daily Claude Code users. Most report a full workday of Sonnet-driven agent loops before the 5-hour window bites; heavy Opus use hits the weekly cap by day 4–5.
Is Claude Max (was Max 5x) worth it? →View plan ↗

Anthropic · Claude Code

Claude Ultra (was Max 20x)

$200/mo

Est. value score (editorial estimate)

86
Official limits
Max 20x was renamed Ultra in Nov 2026; existing subscribers moved over automatically at the same price. No 5-hour limit for verified accounts (new accounts keep a temporary session limit until Anthropic verifies them automatically). 10x Pro usage every week; Anthropic says the weekly limit itself is unchanged (the old "20x" was per session). Shared across chat and Claude Code. Separate weekly Fable allowance (up to half your weekly limit). Includes $200/month in Claude Platform API credits and priority access at peak times. Billed monthly only. Claude Duet (US research-preview waitlist on Pro and Max) shares these limits, and its background check-ins count toward them.
Source
claude.com/pricing · verified 2026-11-02
Crowd-reported burn
Parallel-agent crowd's favorite. The weekly cap was already the real ceiling with 2–3 concurrent agents; now that verified accounts have no 5-hour limit, it's the only one.
Is Claude Ultra (was Max 20x) worth it? →View plan ↗

OpenAI · Codex

ChatGPT Pro 200

$200/mo

Est. value score (editorial estimate)

80
Official limits
No 5-hour Codex limit; weekly limits apply. More usage than Pro 100; no Ultrafast mode (that's Pro 500 only). GPT-6 Astra included in Codex allowance. OpenAI does not publish exact Pro allowances. On Oct 29, 2026 OpenAI canceled the planned Pro 200 allowance cut: subscribers who had the higher Work and Codex allowance keep it, and the lower allowance (new subscribers since Sep 29) doubled to match, so all Pro 200 subscriptions now get the same allowance. The lower weekly GPT-6 Pro message limit was also raised to match the higher one. Billed monthly only.
Source
help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers · verified 2026-11-02Note: Exact Pro allowances aren't published. Pro 200 is open to new subscribers again with the same allowance. Heads up: starting in December 2026, OpenAI says Plus, Pro and Business usage will be one token-based allowance shared across ChatGPT and Codex (Instant chat excluded); pool sizes and start day not published yet.
Crowd-reported burn
Very hard to exhaust for a single dev under the old allowance, which every Pro 200 subscriber now has again after OpenAI canceled the cut on Oct 29, 2026.
Is ChatGPT Pro 200 worth it? →View plan ↗

OpenAI · Codex

ChatGPT Pro 100

$100/mo

Est. value score (editorial estimate)

80
Official limits
Work and Codex allowance doubled on Oct 29, 2026. No 5-hour Codex limit; weekly limits apply. GPT-6 Astra included in Codex allowance. Exact allowance not published.
Source
help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers · verified 2026-11-02Note: Exact allowance not published by OpenAI. Heads up: starting in December 2026, OpenAI says Plus, Pro and Business usage will be one token-based allowance shared across ChatGPT and Codex (Instant chat excluded); pool sizes and start day not published yet.
Crowd-reported burn
No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is ChatGPT Pro 100 worth it? →View plan ↗

OpenAI · Codex

ChatGPT Plus

$20/mo

Est. value score (editorial estimate)

80
Official limits
Codex on web, CLI, IDE and iOS, with cloud code review. Per 5-hour window (estimated by OpenAI): 15–160 GPT-6.1 Sol, 35–285 GPT-6.1 Terra or 350–3,000 GPT-6 Luna local messages; GPT-6 Astra limited (5–45). Weekly limits may also apply; extra credits can be bought.
Source
learn.chatgpt.com/docs/pricing · verified 2026-11-02Note: Heads up: starting in December 2026, OpenAI says Plus, Pro and Business usage will be one token-based allowance shared across ChatGPT and Codex (Instant chat excluded); pool sizes and start day not published yet.
Crowd-reported burn
Best $20 entry for agentic coding right now per submissions: several focused sessions a week before the weekly wall.
Is ChatGPT Plus worth it? →View plan ↗

Google · Gemini / Jules / Antigravity

Google AI Pro

$19.99/mo

Est. value score (editorial estimate)

76
Official limits
4x Free usage limits in the Gemini app (refresh every 5 hours up to a weekly limit), Gemini 4 Argon access, 1M-token context, entry rate limits for the Antigravity agent. Bundled with 5 TB storage.
Source
gemini.google/subscriptions/ · verified 2026-11-02Note: Google's subscription page no longer states specific Gemini CLI / Code Assist request limits, and as of Nov 2, 2026 it no longer lists Jules.
Crowd-reported burn
Generous request counts and a huge context window. Quality on long agent loops varies by task; strong for big-codebase reading.
Is Google AI Pro worth it? →View plan ↗

GitHub · GitHub Copilot

Copilot Pro+

$39/mo

Est. value score (editorial estimate)

74
Official limits
$70/mo of GitHub AI Credits ($39 base + ~$31 variable flex allotment). Same features as Pro; resets monthly; paid overage optional.
Source
github.com/features/copilot/plans · verified 2026-11-02Note: Flex allotment is variable and "may change over time" per GitHub.
Crowd-reported burn
No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is Copilot Pro+ worth it? →View plan ↗

Anthropic · Claude Code

Claude Pro

$20/mo

Est. value score (editorial estimate)

72
Official limits
$20/mo billed monthly ($17/mo billed annually). Claude Code included. At least 5x Free usage per rolling 5-hour session, plus weekly limits, shared across chat and Claude Code. Opus included; Fable only via usage credits. Usage credits at API rates once you hit the limit. Claude Duet (US research-preview waitlist on Pro and Max) shares these limits, and its background check-ins count toward them.
Source
claude.com/pricing · verified 2026-11-02
Crowd-reported burn
Great for 1–2 focused hours per session; agent-heavy refactors hit the 5-hour limit in ~60–90 minutes.
Is Claude Pro worth it? →View plan ↗

OpenAI · Codex

ChatGPT Pro 500

$500/mo

Est. value score (editorial estimate)

72
Official limits
Highest included usage of OpenAI's three Pro plans. Work and Codex allowance doubled on Oct 29, 2026. The only Pro plan with Ultrafast mode for GPT-6 Astra and GPT-6.1 Sol, which burns included usage at 8x the Standard rate (then credits at 6x). No 5-hour Codex limit; weekly limits apply. Exact allowance not published. Billed monthly only.
Source
help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers · verified 2026-11-02Note: Exact allowance not published by OpenAI; Ultrafast use drains the allowance 8x faster. Heads up: starting in December 2026, OpenAI says Plus, Pro and Business usage will be one token-based allowance shared across ChatGPT and Codex (Instant chat excluded); pool sizes and start day not published yet.
Crowd-reported burn
New tier launched Oct 2026. No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is ChatGPT Pro 500 worth it? →View plan ↗

GitHub · GitHub Copilot

Copilot Pro

$10/mo

Est. value score (editorial estimate)

70
Official limits
$15/mo of GitHub AI Credits ($10 base + ~$5 variable flex allotment; 1 credit = $0.01). Credits are used by chat, agent mode, cloud agent, CLI and code review; completions are unlimited. Resets monthly; paid overage optional.
Source
github.com/features/copilot/plans · verified 2026-11-02Note: Flex allotment is variable and "may change over time" per GitHub.
Crowd-reported burn
No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is Copilot Pro worth it? →View plan ↗

GitHub · GitHub Copilot

Copilot Max

$100/mo

Est. value score (editorial estimate)

70
Official limits
$200/mo of GitHub AI Credits ($100 base + $100 variable flex allotment). Same features as Pro+; resets monthly; paid overage optional. Built for sustained agent-driven workflows.
Source
github.com/features/copilot/plans · verified 2026-11-02Note: Flex allotment is variable and "may change over time" per GitHub.
Crowd-reported burn
No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is Copilot Max worth it? →View plan ↗

Cognition · Windsurf

Windsurf Pro

$20/mo

Est. value score (editorial estimate)

68
Official limits
Usage allowance that refreshes daily and weekly (no credits). Frontier OpenAI, Claude, Gemini and SpaceXAI models plus open-source models, free use of Cognition's SWE-2 model, cloud agents (Devin Cloud), up to 10 concurrent sessions; extra usage at API pricing. Mar 2026 launch figures: ~7–27 premium-plus or 8–101 premium messages/day.
Source
devin.ai/pricing · verified 2026-11-02Note: Price verified on devin.ai/pricing (windsurf.com/pricing redirects there). Per-day message ranges are from the Mar 2026 pricing announcement and may be stale.
Crowd-reported burn
No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is Windsurf Pro worth it? →View plan ↗

X / Cursor · Cursor

Cursor Ultra

$200/mo

Est. value score (editorial estimate)

66
Official limits
50x the Plus usage pool (weekly refresh), priority access, earliest access to new products and models. Third-party models billed separately at API list price. Legacy Cursor Ultra ($200) subscribers keep a separate monthly Other Models pool for third-party models; switching to The X Subscription is permanent and drops it.
Source
cursor.com/pricing · verified 2026-11-02
Crowd-reported burn
Plan was restructured; Pro-era reports (frontier models in the pool) no longer apply. Re-collecting data.
Is Cursor Ultra worth it? →View plan ↗

Cognition · Windsurf

Windsurf Max

$200/mo

Est. value score (editorial estimate)

64
Official limits
Significantly higher daily/weekly quotas than Pro, plus priority support. Mar 2026 launch figures: ~42–170 premium-plus or 47–631 premium messages/day.
Source
devin.ai/pricing · verified 2026-11-02Note: Per-day message ranges are from the Mar 2026 announcement and may be stale.
Crowd-reported burn
No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is Windsurf Max worth it? →View plan ↗

X / Cursor · Cursor

Cursor Super

$100/mo

Est. value score (editorial estimate)

64
Official limits
15x the Plus usage pool (weekly refresh) with priority model access at peak times. Third-party models billed separately at API list price.
Source
cursor.com/pricing · verified 2026-11-02
Crowd-reported burn
No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is Cursor Super worth it? →View plan ↗

X / Cursor · Cursor

Cursor Premium

$30/mo

Est. value score (editorial estimate)

62
Official limits
4x the Plus usage pool, shared across Cursor and Grok, refreshing weekly. Includes Grok models plus cloud agents, Bugbot code review and Origin hosting. Third-party models (Claude, GPT, etc.) are NOT included; they're billed separately at API list price.
Source
cursor.com/pricing · verified 2026-11-02
Crowd-reported burn
New tier: replaced Cursor Pro. Old Pro-era crowd data no longer applies. Value depends heavily on whether you're happy on Grok models; third-party model use is pay-as-you-go.
Is Cursor Premium worth it? →View plan ↗

X / Cursor · Cursor

Cursor Plus

$8/mo

Est. value score (editorial estimate)

60
Official limits
Basic usage pool (the base unit for Premium 4x, Super 15x, Ultra 50x), shared across Cursor and Grok, refreshing weekly. Grok models only: third-party models (Claude, GPT, etc.) are not available on this plan. No cloud agents, Origin code hosting or Bugbot reviews. Includes Grok Chat, Voice, Imagine and Build, plus Grok API access (billed separately).
Source
cursor.com/pricing · verified 2026-11-02Note: Cursor doesn't publish the size of the Plus pool, only the multiples for higher tiers.
Crowd-reported burn
Cheapest paid way into Cursor's agent, but Grok-only and the smallest pool. No crowd submissions yet for this plan. Score is a provisional estimate from official limits only.
Is Cursor Plus worth it? →View plan ↗

OpenRouter · Any agent (BYOK)

Pay-as-you-go API

Usage-based

Est. value score (editorial estimate)

60
Official limits
No plan cap. Inference at each provider's list price across 600+ models, plus a 5.5% fee on credit purchases (Standard). No plan-based rate limits on paid models; free models capped at 20 req/min and 50 req/day (1,000/day after $10 of credits).
Source
openrouter.ai/pricing · verified 2026-11-02
Crowd-reported burn
Best for bursty or experimental use and open-weight models. Heavy daily agent use usually costs more than a flat plan.
Is Pay-as-you-go API worth it? →View plan ↗