PlanetGPU
AI coding agents · verified 2026-06-14

AI coding agent pricing, quotas, and buyer value

Compare Codex, Claude Code, Cursor, Windsurf, Kimi, GLM, MiniMax, Qwen, and OpenCode by quota transparency, overage behavior, supported models, BYOK, and where work actually runs.

Read source policyCompare GPU pricesOfficial sources preferredEstimates labeled
AI coding agents and developer tooling plans represented as connected compute and workflow blocks.
Plans tracked
17
High transparency
6
BYOK or partial
14
plans with key flexibility
Estimate caveat
labeled
API value is not entitlement

Featured buyer checkpoints

Estimates are not vendor entitlements.
OpenAI Codex / ChatGPT
Pro 20x
$200/mo
API value: ~$14,000
Claude Code / Claude
Max 20x
$200/mo
API value: ~$8,000
Cursor
Pro
$20/mo
API value: ~$20
Windsurf / Devin Desktop
Pro
$20/mo
API value: ~$20
OpenCode Go
Go
$10/mo
API value: ~$60

Compare plans

17 of 17 plans shown. Filter by budget, surface, and quota transparency.

Sort

OpenCode Go

high transparency
Anomaly · Go · verified 2026-06-14
CLIModel gatewayBYOK yeshybrid
$10/mo
API value ~$60
Included usage / quota

$5 first month, then $10/month. Published model-specific requests per 5 hours include GLM-5.1, Qwen3.7, Kimi K2.7 Code, MiMo, DeepSeek, and MiniMax M3.

Best for

Budget-conscious developers who want cheap access to open coding models.

Overage

Top up credit if needed.

Supported models
GLM-5.1Kimi K2.7 CodeKimi K2.6Qwen3.7 MaxQwen3.7 PlusMiniMax M3+2
Surfaces

OpenCode, Any compatible agent

Low-cost model gateway with generous 5-hour request limits; exact dollar equivalent varies by selected model.

Z.ai GLM Coding Plan

low transparency
Z.ai · GLM Coding Plan · verified 2026-06-14
Model gatewayCLIIDEBYOK yeshybrid
$18/mo
API value ~$18
Included usage / quota

Use GLM models like GLM-5.1 and GLM-5-Turbo in Claude Code, Cursor, Cline, OpenCode, Kilo Code, and related tools.

Best for

Budget users testing GLM as a cheaper backend for existing coding-agent shells.

Overage

Not clearly exposed in crawlable official pricing text.

Supported models
GLM-5.1GLM-5-Turbo
Surfaces

Claude Code, Cursor, Cline, Trae, Kilo Code, OpenCode, API key

Official page confirms plans from 18/month in search-visible text, but structured quota details were not available.

OpenAI Codex / ChatGPT

high transparency
OpenAI · Plus · verified 2026-06-14
CLIIDEBrowser/cloudBYOK partialhybrid
$20/mo
API value ~$20
Included usage / quota

Codex included with Plus. Published 5-hour examples: GPT-5.5 local messages 15-80, GPT-5.4 20-100, GPT-5.4 mini 60-350. Weekly limits may apply.

Best for

Solo developers doing a few focused coding sessions each week.

Overage

ChatGPT credits after included limits; Codex credit burn maps to token-based rate card.

Supported models
GPT-5.5GPT-5.4GPT-5.4 miniGPT-5.3-Codex
Surfaces

Web, CLI, VS Code IDE extension, iOS, Slack, GitHub review

Included usage can exceed direct API value for active coding, but official entitlement is quota/credits, not dollars.

Claude Code / Claude

medium transparency
Anthropic · Pro · verified 2026-06-14
CLIIDEBrowser/cloudBYOK partialhybrid
$20/mo
API value ~$20
Included usage / quota

Claude Code included. Usage is shared across claude.ai, Claude Desktop/mobile, and Claude Code. Session limits reset every five hours and weekly limits apply.

Best for

Claude-first developers with moderate terminal-agent usage.

Overage

Optional usage credits after plan limits, billed at standard API rates.

Supported models
Claude OpusClaude SonnetClaude Haiku
Surfaces

Terminal, VS Code, Cursor and VS Code forks, JetBrains, Web, Desktop, Mobile

Quota is not published as tokens or dollars; usage varies by model, context, effort, and feature.

Cursor

high transparency
Anysphere · Pro · verified 2026-06-14
IDECLIBrowser/cloudBYOK partialhybrid
$20/mo
API value ~$20
Included usage / quota

Unlimited Tab and Auto mode plus $20/month of frontier model usage at API pricing.

Best for

IDE-first developers who want a predictable $20 plan plus Auto mode.

Overage

On-demand frontier model usage at cost after included usage; spend limit available.

Supported models
OpenAIAnthropicGooglexAICursor custom models
Surfaces

Cursor IDE, Agents, Cloud agents, CLI, Bugbot

Cursor explicitly describes Pro as $20 of frontier model usage at API pricing, plus unlimited Auto/Tab.

Windsurf / Devin Desktop

low transparency
Cognition · Pro · verified 2026-06-14
IDEBrowser/cloudBYOK unknownhybrid
$20/mo
API value ~$20
Included usage / quota

Increased quotas, full model availability, free SWE 1.6 and leading open-source models, Devin Cloud access.

Best for

Developers who prefer the Windsurf/Devin IDE workflow over raw CLI agents.

Overage

Purchase extra usage at API pricing.

Supported models
SWE-1.6SWE-1.6 FastSWE-1.5OpenAIClaudeGemini+1
Surfaces

Devin Desktop, Cascade, Devin Cloud

Quota language is qualitative; official page does not expose a token/credit pool for self-serve plans.

MiniMax Code / Token Plan

medium transparency
MiniMax · Token Plan · verified 2026-06-14
IDEModel gatewayBrowser/cloudBYOK partialhybrid
$20/mo
API value ~$200
Included usage / quota

Official homepage positions Token Plan as revamped with 2x tokens, up to 12.5B tokens/month, MiniMax M3, 1M context, and frontier coding capabilities.

Best for

Developers exploring high-throughput open/Chinese model coding at low subscription cost.

Overage

Token/API plan details available through MiniMax platform.

Supported models
MiniMax M3MiniMax M2.7MiniMax M2.5
Surfaces

MiniMax Code, Desktop, API, MiniMax Agent

Sources
Official

Vendor marketing claims $20 equals 10x Claude Pro throughput; keep as claim, not normalized entitlement.

Cursor

medium transparency
Anysphere · Pro+ · verified 2026-06-14
IDECLIBrowser/cloudBYOK partialhybrid
$60/mo
API value ~$60
Included usage / quota

Recommended by Cursor for daily agent users. Public secondary summaries report 3x Pro usage.

Best for

Daily Cursor agent users who outgrow Pro.

Overage

On-demand model usage after included usage.

Supported models
OpenAIAnthropicGooglexAICursor custom models
Surfaces

Cursor IDE, Agents, Cloud agents, CLI, Bugbot

Inferred from public Pro+ positioning and secondary summaries; verify in billing dashboard before publishing as hard value.

OpenAI Codex / ChatGPT

high transparency
OpenAI · Pro 5x · verified 2026-06-14
CLIIDEBrowser/cloudBYOK partialhybrid
$100/mo
API value ~$100
Included usage / quota

5x Plus Codex usage. Published 5-hour examples: GPT-5.5 local messages 80-400, GPT-5.4 100-500, GPT-5.4 mini 300-1750.

Best for

Daily professional Codex users who need longer sessions.

Overage

ChatGPT credits after included limits; token-based Codex rate card applies.

Supported models
GPT-5.5GPT-5.4GPT-5.4 miniGPT-5.3-CodexGPT-5.3-Codex-Spark preview
Surfaces

Web, CLI, VS Code IDE extension, iOS, Slack, GitHub review

Likely above direct API value under heavy use, but variable by model, context, output, and fast mode.

Claude Code / Claude

medium transparency
Anthropic · Max 5x · verified 2026-06-14
CLIIDEBrowser/cloudBYOK partialhybrid
$100/mo
API value ~$100
Included usage / quota

5x more usage per session than Pro, with weekly all-model and Sonnet-specific limits.

Best for

Frequent Claude Code users who hit Pro caps.

Overage

Optional usage credits after plan limits, billed at standard API rates.

Supported models
Claude OpusClaude SonnetClaude Haiku
Surfaces

Terminal, VS Code, Cursor and VS Code forks, JetBrains, Web, Desktop, Mobile

Sources
HelpHelp

Likely valuable for heavy usage, but not directly convertible because Anthropic does not publish exact token quota.

Windsurf / Devin Desktop

low transparency
Cognition · Teams · verified 2026-06-14
IDEBrowser/cloudBYOK unknownhybrid
$120/mo
API value ~$120
Included usage / quota

$80/month team plan plus $40/month per full dev seat. Full users include their own quota and Devin Desktop access.

Best for

Small teams that want shared Devin Desktop and cloud agent workflow.

Overage

Purchase extra usage at API pricing.

Supported models
SWE-1.6SWE-1.6 FastSWE-1.5OpenAIClaudeGemini+1
Surfaces

Devin Desktop, Cascade, Devin Cloud, Team collaboration

Sources
Official

Team price includes base plan plus one full dev seat; quota not mapped to API dollars.

OpenAI Codex / ChatGPT

high transparency
OpenAI · Pro 20x · verified 2026-06-14
CLIIDEBrowser/cloudBYOK partialhybrid
$200/mo
API value ~$14,000
Included usage / quota

20x Plus Codex usage. Published 5-hour examples: GPT-5.5 local messages 300-1600, GPT-5.4 400-2000, GPT-5.4 mini 1200-7000.

Best for

Agent-heavy developers running Codex as a primary coding workflow.

Overage

ChatGPT credits after included limits; token-based Codex rate card applies.

Supported models
GPT-5.5GPT-5.4GPT-5.4 miniGPT-5.3-CodexGPT-5.3-Codex-Spark preview
Surfaces

Web, CLI, VS Code IDE extension, iOS, Slack, GitHub review

High-utilization secondary estimate, not an official OpenAI entitlement. Official source publishes usage windows and credit rates, not a dollar value.

Claude Code / Claude

medium transparency
Anthropic · Max 20x · verified 2026-06-14
CLIIDEBrowser/cloudBYOK partialhybrid
$200/mo
API value ~$8,000
Included usage / quota

20x more usage per session than Pro, with weekly all-model and Sonnet-specific limits.

Best for

Claude-centric power users who need the largest individual Claude Code tier.

Overage

Optional usage credits after plan limits, billed at standard API rates.

Supported models
Claude OpusClaude SonnetClaude Haiku
Surfaces

Terminal, VS Code, Cursor and VS Code forks, JetBrains, Web, Desktop, Mobile

Secondary high-utilization estimate; Anthropic publishes multipliers and limits, not a dollar equivalent.

Cursor

medium transparency
Anysphere · Ultra · verified 2026-06-14
IDECLIBrowser/cloudBYOK partialhybrid
$200/mo
API value ~$200
Included usage / quota

Recommended by Cursor for agent power users. Secondary summaries report 20x Pro usage.

Best for

Cursor power users who want minimum friction inside one IDE.

Overage

On-demand model usage after included usage.

Supported models
OpenAIAnthropicGooglexAICursor custom models
Surfaces

Cursor IDE, Agents, Cloud agents, CLI, Bugbot

Likely close to included frontier-model credit value plus Auto/Tab, but exact value should be verified in account UI.

Windsurf / Devin Desktop

low transparency
Cognition · Max · verified 2026-06-14
IDEBrowser/cloudBYOK unknownhybrid
$200/mo
API value ~$400
Included usage / quota

Everything in Pro plus significantly higher quotas.

Best for

Power users already committed to Devin Desktop/Cascade.

Overage

Purchase extra usage at API pricing.

Supported models
SWE-1.6SWE-1.6 FastSWE-1.5OpenAIClaudeGemini+1
Surfaces

Devin Desktop, Cascade, Devin Cloud

Unverified buyer estimate. Official page says significantly higher quotas but does not publish a token/dollar mapping.

Kimi / Moonshot

medium transparency
Moonshot AI · Kimi Code / API · verified 2026-06-14
CLIBrowser/cloudModel gatewayBYOK yeshybrid
Usage based
API value Usage based
Included usage / quota

Kimi Code CLI is open-source; model usage is paid through Kimi API or app/account plans. Kimi K2.7 Code is positioned as the strongest coding model.

Best for

Cost-sensitive developers willing to route coding agents through Kimi models.

Overage

Token-based API pricing by model; see Kimi platform pricing pages.

Supported models
Kimi K2.7 CodeKimi K2.6Kimi K2.5Moonshot V1
Surfaces

Kimi web/app, Kimi Code CLI, VS Code extension, ACP-compatible IDEs, API

Usage-based rather than a comparable monthly bundled subscription in this catalog row.

Qwen Code

high transparency
Alibaba / Qwen · Qwen Code + BYOK · verified 2026-06-14
CLIIDEModel gatewayBYOK yeslocal
Usage based
API value Usage based
Included usage / quota

Open-source agent. Qwen OAuth free tier was discontinued on April 15, 2026; use Alibaba Cloud Coding Plan, OpenRouter, Fireworks, or BYOK.

Best for

Developers who want an open-source CLI agent and direct model/provider control.

Overage

Provider/API billing through Alibaba Cloud Model Studio, OpenRouter, Fireworks, or compatible APIs.

Supported models
Qwen3-CoderQwen3.6 PlusQwen3.5 PlusOpenAI-compatible modelsAnthropicGemini
Surfaces

Terminal, VS Code, Zed, JetBrains, Claude Code proxy, Cline

The tool is free/open-source; value depends on the selected API provider or coding plan.

Source policy

PlanetGPU prioritizes official pricing, docs, and help-center pages. Secondary sources are used only where official sources are vague, and those values are marked as estimates or low-confidence buyer-value signals. The v1 catalog is curated static data; scraper-backed refresh and source snapshots are intentionally deferred.