AI Coding Plan Comparison
Updated 12 July 2026 · first published 4 July 2026
Compare major AI coding plans with verifiable evidence. Pricing, limits, and model scores in one normalised table. Built from provider docs and explicit conversion rules for fair comparison.
- Every update is tied to public evidence
- Prompt / Request conversion rules are explicit
- Fixed calculation logic keeps comparisons consistent
Methodology
Collection
Plan, limit, and model data are collected from public provider sources.
Normalisation
Prompt / request / period formats are normalised into one comparison model.
Conversion rules
Request / Prompt = 15 and Weekly × 4 = Monthly are applied explicitly.
Display policy
Official values and estimated values are clearly separated in the UI.
Update Feed
coding-plan-comparison research page launched
Published evidence-first AI coding plan comparison as a FinOps LLM research article.
Xiaomi MiMO quota increased
Xiaomi MiMO increased the available usage limits, and the comparison table has been updated.
Alibaba sunsets Coding Plan, adds Token Plan (Team Edition)
Alibaba closed the previous Coding Plan and introduced Token Plan (Team Edition).
Coding Plan Comparison List
| Provider ↕ | Plan ↕ | Price ↕ | Uses | Prompt / mo ↕ | Request / mo ↕ | Prompt / $1 ↕ | Request / $1 ↕ | Models | Intel ↕ | Coding ↕ | Agentic ↕ |
|---|---|---|---|---|---|---|---|---|---|---|---|
| GLM | Lite | $10 | Coding + Agent | 160 | 2,400 | 16 | 240 | GLM-5.1 (×3), GLM-5-Turbo (×3), GLM-4.7 (×1) | 47 | 37 | 66 |
| GLM | Pro | $30 | Coding + Agent | 267 | 4,000 | 8.9 | 133 | GLM-5.1 (×3), GLM-5-Turbo (×3), GLM-5 (×3) | 50 | 44 | 63 |
| GLM | Max | $80 | Coding + Agent | 400 | 6,000 | 5 | 75 | GLM-5.1 (×3), GLM-5-Turbo (×3), GLM-5 (×3) | 50 | 44 | 63 |
| MiniMax | Starter | $10 | Coding + Agent | 400 | 6,000 | 40 | 600 | MiniMax-M2.7 (×1) | 50 | 42 | 61 |
| MiniMax | Plus | $20 | Coding + Agent | 600 | 9,000 | 30 | 450 | MiniMax-M2.7 (×1) | 50 | 42 | 61 |
| MiniMax | Max | $45 | Coding + Agent | 889 | 13,333 | 19.8 | 296 | MiniMax-M2.7 (×1) | 50 | 42 | 61 |
| MiniMax | Plus-HighSpeed | $40 | Coding + Agent | 300 | 4,500 | 7.5 | 112.5 | MiniMax-M2.7-highspeed (×1) | 50 | 42 | 61 |
| MiniMax | Max-HighSpeed | $80 | Coding + Agent | 500 | 7,500 | 6.25 | 93.75 | MiniMax-M2.7-highspeed (×1) | 50 | 42 | 61 |
| MiniMax | Ultra-HighSpeed | $150 | Coding + Agent | 533 | 8,000 | 3.55 | 53.3 | MiniMax-M2.7 (×1) | 50 | 42 | 61 |
| BytePlus | Lite | $10 | Coding + Agent | 120 | 1,800 | 12 | 180 | Seed-Code (×1), GLM-4.7 (×1), DeepSeek-V3.2 (×1), kimi-k2.5 (×1) | 34 | 31 | 36 |
| BytePlus | Pro | $50 | Coding + Agent | 120 | 1,800 | 2.4 | 36 | Seed-Code (×1), GLM-4.7 (×1), DeepSeek-V3.2 (×1), kimi-k2.5 (×1) | 34 | 31 | 36 |
| StepFun | Flash Mini | $6.99 | Coding | 229 | 3,433 | 32.8 | 491 | step-3.5-flash (×1) | 38 | 32 | 52 |
| StepFun | Flash Plus | $9.99 | Coding | 641 | 9,610 | 64.2 | 962 | step-3.5-flash (×1) | 38 | 32 | 52 |
| StepFun | Flash Pro | $29 | Coding | 828 | 12,414 | 28.6 | 428 | step-3.5-flash (×1) | 38 | 32 | 52 |
| StepFun | Flash Max | $99 | Coding | 808 | 12,121 | 8.16 | 122.4 | step-3.5-flash (×1) | 38 | 32 | 52 |
| KwaiKAT | Mini | $5 | Coding | 1,075 | 16,128 | 215 | 3,226 | KAT-Coder-Pro V2 (×1) | 44 | 46 | 51 |
| KwaiKAT | Starter | $10 | Coding | 1,344 | 20,160 | 134.4 | 2,016 | KAT-Coder-Pro V2 (×1) | 44 | 46 | 51 |
| KwaiKAT | Pro | $20 | Coding | 2,016 | 30,240 | 100.8 | 1,512 | KAT-Coder-Pro V2 (×1) | 44 | 46 | 51 |
| KwaiKAT | Max | $50 | Coding | 2,688 | 40,320 | 53.76 | 806.4 | KAT-Coder-Pro V2 (×1) | 44 | 46 | 51 |
| OpenCode | Go | $10 | Coding + Agent | 667 | 10,000 | 66.7 | 1,000 | MiniMax-M2.5 (×1), M2.7 (×1.4), MiMo-V2-Pro (×16), kimi-k2.5 (×11), GLM-5 (×17) | 50 | 44 | 63 |
| Xiaomi MiMO Sign up - both get $2 ↗ |
Lite | $6 | Coding + Agent | 380 | 5,693 | 63.3 | 949 | MiMo-V2-Pro (×1), MiMo-V2-Omni (×1) | 49 | 41 | 63 |
| Xiaomi MiMO Sign up - both get $2 ↗ |
Standard | $16 | Coding + Agent | 382 | 5,729 | 23.9 | 578 | MiMo-V2-Pro (×1), MiMo-V2-Omni (×1) | 49 | 41 | 63 |
| Xiaomi MiMO Sign up - both get $2 ↗ |
Pro | $50 | Coding + Agent | 422 | 6,333 | 8.44 | 126.7 | MiMo-V2-Pro (×1), MiMo-V2-Omni (×1) | 49 | 41 | 63 |
| Xiaomi MiMO Sign up - both get $2 ↗ |
Max | $100 | Coding + Agent | 456 | 6,833 | 4.56 | 68.3 | MiMo-V2-Pro (×1), MiMo-V2-Omni (×1) | 49 | 41 | 63 |
| Alibaba | Standard | $30 | Coding + Agent | 17 | 262 | 0.57 | 8.73 | qwen3.6-plus (×1), deepseek-v3.2 (×1), GLM-5 (×1), MiniMax-M2.5 (×1) | 50 | 43 | 62 |
| Alibaba | Advanced | $100 | Coding + Agent | 21 | 314 | 0.21 | 3.14 | qwen3.6-plus (×1), deepseek-v3.2 (×1), GLM-5 (×1), MiniMax-M2.5 (×1) | 50 | 43 | 62 |
| Alibaba | Premium | $200 | Coding + Agent | 26 | 393 | 0.13 | 1.97 | qwen3.6-plus (×1), deepseek-v3.2 (×1), GLM-5 (×1), MiniMax-M2.5 (×1) | 50 | 43 | 62 |
Values not explicitly provided by providers are shown in gray.
Many providers limit either Prompt or Request quotas. We convert dynamically using the Request / Prompt rate control above (default: 15).
If the highest official limit is weekly, monthly is estimated as Weekly × 4.
Artificial Analysis Score is measured by artificialanalysis.ai. See their site for details.
Some links may include referral codes.
Plan details can change frequently. Always verify the latest information on official provider pages.
Practical Guide
How to choose a coding plan
Price alone is not enough. Match each plan to your working style first. Start with Agent policy, limit type, and contract horizon.
1) Check Agent policy first
Some providers or plans restrict Agent usage. If Agent workflows matter, verify terms and the Uses column (Coding / Agent) before subscribing.
2) Prompt limits vs Request limits
For manual chat-heavy usage, batching work into fewer turns often aligns better with Prompt-style limits. For automation and Agent-heavy workflows, backend request volume can grow quickly, so prioritise plans with sufficient Request headroom.
3) Prefer monthly, use annual when early terms are superior
Plan revisions and enrolment changes are increasingly common. Monthly gives flexibility, while annual can be effective when launch-period terms are clearly advantageous.
4) Speed (token/s) is a real productivity factor
With the same quota, slower responses still cost more working time. We plan to add token-per-second metrics in a future update.
Additional checks worth adding
- Separate Provided from Estimate values (estimates can shift)
- Confirm same-class model availability (none means unavailable)
- Match reset cycle (5h / daily / monthly) to your actual usage window
Detailed cost comparison: See Cursor vs Copilot vs Claude Code Cost for per-tier pricing, credit value, and total cost breakdowns by subscription level.
Want this applied to your own LLM spend? FinOps LLM runs a free audit of your AI costs and shows where the savings are. Book free audit →