# AI coding plans compared

AI coding plans cost $8 to $500 a month across the 24 plans and 8 products we read from provider pages between October 16, 2026 and October 26, 2026. The cheapest is ChatGPT Go. The strongest model any plan includes, by BenchLM's coding score, is Claude Sonnet 5.5 (84.5, #1 of 154), from Claude Pro at $20. Most providers publish multiples or credits rather than a token cap, so the table shows each plan's meter as published.

Prices read from provider pages; coding scores from the BenchLM coding leaderboard; API rates from the pricing registry (updated October 26, 2026).

## Plans

| Plan | Price | Coding tool | Strongest included model (coding score) | Limits | Plan price at that model's API rates | Source |
|---|---|---|---|---|---|---|
| ChatGPT Go | $8/month | Codex | GPT-5.6 Terra (61.8, #16) | More messages with tools, more uploads, more image creation, more voice chats and longer memory than Free; limited Codex access, allowance not itemized on the page. (5-hour window) | 0.7M output tokens | [ChatGPT: Codex pricing and plans](https://learn.chatgpt.com/docs/pricing), read 2026-10-16 |
| OpenCode Go | $10/month | OpenCode (single-member workspaces) | Gemini 4 Argon (73.2, #6) | Per-model monthly allowances in dollars of usage; 20% of a model's monthly allowance is available per five hours and 50% per week. (5-hour window, weekly cap) | 1M output tokens | [OpenCode Go docs](https://opencode.ai/docs/go), read 2026-10-26 |
| GLM Coding Plan Lite | $18/month ($12.6/month billed annually) | Bring your own tool: Claude Code, Codex, OpenCode, Cline and 20+ agent tools, plus Z.ai ZCode | GLM-5.3 (56.2, #29) | 10,000 credits a week and 2,000 credits per five hours; credits are charged per token, so the token allowance depends on the model and the cache hit rate. (5-hour window, weekly cap) | No list price | [GLM Coding Plan](https://z.ai/subscribe), read 2026-10-26 |
| Kimi Plus | $19/month ($15/month billed annually) | Kimi Code (CLI, desktop and IDE plugins) | Kimi K3 (59, #22) | Kimi Code with a rolling five-hour limit and a monthly total quota, and no weekly limit; agent credits come from one monthly pool. Moonshot does not publish the quota in numbers. (5-hour window, no weekly cap) | 1.3M output tokens | [Kimi membership pricing](https://www.kimi.com/membership/pricing), read 2026-10-26 |
| Google AI Pro | $19.99/month | Jules and Google Antigravity (entry rate limits) | Gemini 4 Argon (73.2, #6) | 4x the usage limits of Free. Usage limits in the Gemini app are compute-based (prompt complexity, features used and chat length count) and refresh every 5 hours until the weekly limit is reached. (5-hour window, weekly cap) | 2M output tokens | [Gemini subscriptions (United States)](https://gemini.google/us/subscriptions/?hl=en), read 2026-10-21 |
| ChatGPT Plus | $20/month | Codex | GPT-6 Astra (75.5, #4) | Estimated local Codex messages per five-hour window: 5 to 45 with GPT-6 Astra, 15 to 160 with GPT-6.1 Sol, 15 to 150 with GPT-6 Sol, 350 to 3,000 with GPT-6 Luna; weekly limits may also apply. (5-hour window) | 0.4M output tokens | [ChatGPT: Codex pricing and plans](https://learn.chatgpt.com/docs/pricing), read 2026-10-16 |
| Claude Pro | $20/month ($17/month billed annually) | Claude Code (terminal, IDE, desktop and web) | Claude Sonnet 5.5 (84.5, #1) | At least 5x more usage per five-hour session than Free, plus a weekly limit on paid plans. (5-hour window, weekly cap) | 2M output tokens | [Claude pricing](https://claude.com/pricing), read 2026-10-16 |
| MiniMax M Plan Go | $22/month ($18.33/month billed annually) | MiniMax Code, plus an API key for Claude Code, Codex, Cursor, OpenCode and other tools | MiniMax M3.1 Flash (58.9, #23) | The base tier; MiniMax does not publish the quota in numbers. (Not published) | 36.7M output tokens | [MiniMax M Plan](https://www.minimax.io/m-plan), read 2026-10-26 |
| Kimi Pro | $39/month ($31/month billed annually) | Kimi Code (CLI, desktop and IDE plugins) | Kimi K3 (59, #22) | 2x agent credits. Kimi Code with a rolling five-hour limit and a monthly total quota, and no weekly limit; agent credits come from one monthly pool. Moonshot does not publish the quota in numbers. (5-hour window, no weekly cap) | 2.6M output tokens | [Kimi membership pricing](https://www.kimi.com/membership/pricing), read 2026-10-26 |
| OpenCode Go Plus | $40/month | OpenCode (single-member workspaces) | Gemini 4 Argon (73.2, #6) | Higher per-model allowances than Go across the same models, with the same five-hour (20%) and weekly (50%) shares. (5-hour window, weekly cap) | 4M output tokens | [OpenCode Go docs](https://opencode.ai/docs/go), read 2026-10-26 |
| Alibaba Model Studio Coding Plan Pro | $50/month | API key for Claude Code, OpenCode, Cursor, Codex, Qwen Code, Cline and other tools; interactive use only | Qwen3.7 Plus (40.5, #68) | Up to 6,000 requests per five hours, 45,000 a week and 90,000 a month; Alibaba says a simple task uses 5 to 10 model calls and a complex one 10 to 30 or more. (5-hour window, weekly cap) | No list price | [Alibaba Cloud Coding Plan overview](https://www.alibabacloud.com/help/en/model-studio/coding-plan), read 2026-10-26 |
| MiniMax M Plan Explore | $55/month ($45.83/month billed annually) | MiniMax Code, plus an API key for Claude Code, Codex, Cursor, OpenCode and other tools | MiniMax M3.1 Flash (58.9, #23) | 3x Go usage. (Not published) | 91.7M output tokens | [MiniMax M Plan](https://www.minimax.io/m-plan), read 2026-10-26 |
| GLM Coding Plan Pro | $80/month ($56/month billed annually) | Bring your own tool: Claude Code, Codex, OpenCode, Cline and 20+ agent tools, plus Z.ai ZCode | GLM-5.3 (56.2, #29) | 6x Lite usage: 60,000 credits a week and 12,000 per five hours. (5-hour window, weekly cap) | No list price | [GLM Coding Plan](https://z.ai/subscribe), read 2026-10-26 |
| Kimi Max | $99/month ($79/month billed annually) | Kimi Code (CLI, desktop and IDE plugins) | Kimi K3 (59, #22) | 5x agent credits. Kimi Code with a rolling five-hour limit and a monthly total quota, and no weekly limit; agent credits come from one monthly pool. Moonshot does not publish the quota in numbers. (5-hour window, no weekly cap) | 6.6M output tokens | [Kimi membership pricing](https://www.kimi.com/membership/pricing), read 2026-10-26 |
| Google AI Ultra 5x | $99.99/month | Jules and Google Antigravity (higher rate limits) | Gemini 4 Argon (73.2, #6) | 5x the usage limits of Google AI Pro. Usage limits in the Gemini app are compute-based (prompt complexity, features used and chat length count) and refresh every 5 hours until the weekly limit is reached. (5-hour window, weekly cap) | 10M output tokens | [Gemini subscriptions (United States)](https://gemini.google/us/subscriptions/?hl=en), read 2026-10-21 |
| ChatGPT Pro 100 | $100/month | Codex | GPT-6 Astra (75.5, #4) | The lowest Pro allowance; OpenAI publishes no multiple of Plus. The Codex pricing page says Pro plans currently have no five-hour limit and gives no Pro message estimates. Ultrafast is not included. (Not published) | 2M output tokens | [ChatGPT: Codex pricing and plans](https://learn.chatgpt.com/docs/pricing), read 2026-10-16 |
| Claude Max 5x | $100/month | Claude Code (terminal, IDE, desktop and web) | Claude Sonnet 5.5 (84.5, #1) | 5x more usage per five-hour session than Pro, higher output limits, priority access during high-traffic periods, and a weekly limit across all models. (5-hour window, weekly cap) | 10M output tokens | [Claude Help: What is the Max plan](https://support.claude.com/en/articles/11049741-what-is-the-max-plan), read 2026-10-16 |
| MiniMax M Plan Build | $132/month ($110/month billed annually) | MiniMax Code, plus an API key for Claude Code, Codex, Cursor, OpenCode and other tools | MiniMax M3.1 Flash (58.9, #23) | 7.5x Go usage, with priority access during peak hours. (Not published) | 220M output tokens | [MiniMax M Plan](https://www.minimax.io/m-plan), read 2026-10-26 |
| GLM Coding Plan Max | $168/month ($118/month billed annually) | Bring your own tool: Claude Code, Codex, OpenCode, Cline and 20+ agent tools, plus Z.ai ZCode | GLM-5.3 (56.2, #29) | 14x Lite usage: 140,000 credits a week and 28,000 per five hours. (5-hour window, weekly cap) | No list price | [GLM Coding Plan](https://z.ai/subscribe), read 2026-10-26 |
| Kimi Ultra | $199/month ($159/month billed annually) | Kimi Code (CLI, desktop and IDE plugins) | Kimi K3 (59, #22) | 10x agent credits. Kimi Code with a rolling five-hour limit and a monthly total quota, and no weekly limit; agent credits come from one monthly pool. Moonshot does not publish the quota in numbers. (5-hour window, no weekly cap) | 13.3M output tokens | [Kimi membership pricing](https://www.kimi.com/membership/pricing), read 2026-10-26 |
| Google AI Ultra 20x | $200/month | Jules and Google Antigravity (highest rate limits) | Gemini 4 Argon (73.2, #6) | 20x the usage limits of Google AI Pro. Usage limits in the Gemini app are compute-based (prompt complexity, features used and chat length count) and refresh every 5 hours until the weekly limit is reached. (5-hour window, weekly cap) | 20M output tokens | [Gemini subscriptions (United States)](https://gemini.google/us/subscriptions/?hl=en), read 2026-10-21 |
| ChatGPT Pro 200 | $200/month | Codex | GPT-6 Astra (75.5, #4) | More usage than Pro 100; OpenAI publishes no multiple of Plus. New Pro 200 subscriptions not eligible for grandfathering get a lower allowance than previously offered. Pro plans currently have no five-hour limit. Ultrafast is not included. (Not published) | 4M output tokens | [ChatGPT: Codex pricing and plans](https://learn.chatgpt.com/docs/pricing), read 2026-10-16 |
| Claude Max 20x | $200/month | Claude Code (terminal, IDE, desktop and web) | Claude Sonnet 5.5 (84.5, #1) | 20x more usage per five-hour session than Pro, higher output limits, priority access during high-traffic periods, and a weekly limit across all models. (5-hour window, weekly cap) | 20M output tokens | [Claude Help: What is the Max plan](https://support.claude.com/en/articles/11049741-what-is-the-max-plan), read 2026-10-16 |
| ChatGPT Pro 500 | $500/month | Codex | GPT-6 Astra (75.5, #4) | The highest included usage of the three Pro plans. Ultrafast for GPT-6 Astra and GPT-6.1 Sol uses included usage at 8x the Standard rate, then credits at 6x. Pro plans currently have no five-hour limit. (Not published) | 10M output tokens | [ChatGPT: Codex pricing and plans](https://learn.chatgpt.com/docs/pricing), read 2026-10-16 |

## Frequently asked questions

### What is an AI coding plan?

A monthly subscription that pays for model usage inside a coding tool, instead of per-token API billing. Some plans come with the provider's own tool, such as Claude Code, Codex, Kimi Code or MiniMax Code. Others give you an API key and base URL for tools you already run, such as Claude Code, OpenCode or Cline.

### Which coding plan is cheapest?

ChatGPT Go at $8 a month is the lowest list price in the table. 7 plans cost $20 a month or less. A low price buys different things: dollars of usage per model, credits, requests or an unpublished multiple, so compare the meter as well as the price.

### Which coding plan includes the strongest model?

By BenchLM's coding score, Claude Sonnet 5.5 (84.5) is the strongest model a plan in the table includes. The cheapest plan with it is Claude Pro at $20. A plan that lists a strong model can still give it a small share of the allowance, so check that model's limit on the provider page.

### Are coding plans cheaper than the API?

It depends on your usage, and most providers do not publish plan limits in tokens. The table shows how many output tokens each plan price buys at API list rates. A real API bill also counts input tokens, which coding agents send in volume, so price your own input and output mix before comparing.

### Do these plans work with Claude Code?

Claude Pro and Max include Claude Code. The GLM Coding Plan, MiniMax M Plan and Alibaba's Coding Plan each publish an Anthropic-compatible endpoint, so Claude Code can run their models with the plan's key. Kimi Code and OpenCode Go use their own tools.

## Related

- [Claude pricing plans and Claude Code](https://benchlm.ai/claude/pricing-plans)
- [Claude usage limits](https://benchlm.ai/claude/usage-limits)
- [Codex usage limits](https://benchlm.ai/codex/usage-limits)
- [GitHub Copilot pricing](https://benchlm.ai/github-copilot/pricing)
- [Best LLM for coding](https://benchlm.ai/coding)

Canonical page: https://benchlm.ai/coding-plans