Coding Plan
Flexible subscription plans to meet different development needs.
Coding Plan
With AtlasCloud, deploy state-of-the-art (SOTA) open-source models with a single click
Cut your coding agent costs by
0%SOTA Open-Source Models
Seamless one-click switching between leading models—including DeepSeek, MiniMax, Kimi, and GLM—without the need to subscribe to multiple plans.
Industry-Leading Pricing
Plan rates are priced as low as 50% of pay-per-use rates—35% cheaper than OpenRouter.
Comprehensive Productivity Boost
Gain one-click access to leading tools such as OpenClaw, Claude Code, Codex, and Cursor.
Choose your plan
Your plan runs on a dual-limit system (e.g. 120M / month and 60M / week), pausing automatically once either is reached. The weekly cap resets every Monday at 00:00 UTC with no roll-over; the monthly total stays available for the whole cycle. The generous weekly cap lets heavy users finish a sprint in about two weeks, while the monthly view spares lighter users the pressure of weekly waste. On renewal, all remaining credits are cleared and a new cycle begins from the time of purchase.
Starter
Perfect for individual developers getting started, with reliable daily quota.
- 35%-50% off pay-as-you-go
- Supports 15 top-tier LLM models, continuously updated
- Supports over 20 tools, such as Claude Code, OpenClaw, and more
- Full API access
Lite
Great for daily coding and small projects—best value pick.
- 35%-50% off pay-as-you-go
- Supports 15 top-tier LLM models, continuously updated
- Supports over 20 tools, such as Claude Code, OpenClaw, and more
- Full API access
Plus
The sweet spot between performance and cost — ideal for mid-sized projects.
- 35%-50% off pay-as-you-go
- Supports 15 top-tier LLM models, continuously updated
- Supports over 20 tools, such as Claude Code, OpenClaw, and more
- Full API access
Max
Built for heavier workflows and small-team collaboration.
- 35%-50% off pay-as-you-go
- Supports 15 top-tier LLM models, continuously updated
- Supports over 20 tools, such as Claude Code, OpenClaw, and more
- Full API access
Ultra
For power users with long-running, multi-task workflows.
- 35%-50% off pay-as-you-go
- Supports 15 top-tier LLM models, continuously updated
- Supports over 20 tools, such as Claude Code, OpenClaw, and more
- Full API access
Enterprise
Enterprise-grade, with SLA and custom options.
- 35%-50% off pay-as-you-go
- Supports 15 top-tier LLM models, continuously updated
- Supports over 20 tools, such as Claude Code, OpenClaw, and more
- Full API access
Pay-per-Use Packs
(Valid for 90 days)Supported Models and Multipliers
Points consumed = Input Tokens × Input Multiplier + Output Tokens × Output Multiplier
| Model ID | Input Multiplier | Output Multiplier | Cache Write Multiplier | Cache Read Multiplier | ≈ Savings vs. official pricing |
|---|---|---|---|---|---|
| bytedance/doubao-seed-2.1-turbo-260628 | 1.190 | 5.940 | - | 0.238 | 20% |
| bytedance/doubao-seed-2.1-pro-260628 | 2.380 | 11.880 | - | 0.475 | 20% |
| zai-org/glm-5.2 | 3.700 | 11.620 | - | 0.686 | 20% |
| moonshotai/kimi-k2.7-code | 1.720 | 7.260 | - | 0.345 | 45% |
| minimaxai/minimax-m3 | 0.890 | 3.560 | - | 0.178 | 55% |
| deepseek-ai/deepseek-v4-pro | 2.870 | 5.750 | - | 0.231 | 50% |
| deepseek-ai/deepseek-v4-flash | 0.230 | 0.460 | - | 0.046 | 50% |
| moonshotai/kimi-k2.6 | 1.720 | 7.260 | - | 0.290 | 45% |
| qwen/qwen3.6-plus | 3.300 | 9.900 | 0.660 | 1.980 | 50% |
| zai-org/glm-5.1 | 2.540 | 7.990 | - | 0.472 | 45% |
| minimaxai/minimax-m2.7 | 2.360 | 4.000 | - | 0.109 | 45% |
| minimaxai/minimax-m2.5 | 0.650 | 2.180 | - | 0.109 | 45% |
| zai-org/glm-5 | 1.820 | 5.810 | - | 0.363 | 45% |
| moonshotai/kimi-k2.5 | 1.090 | 5.450 | - | 0.182 | 45% |
| deepseek-ai/deepseek-v3.2 | 0.420 | 0.620 | - | 0.193 | 55% |
Get started in just 3 steps
DocsPurchase a package
Select the appropriate plan based on your usage scenario (development and testing, commercial deployment, or high-concurrency calls).
Use with Coding Tools
Frequently Asked Questions
For more questions, please contact us
Points consumed = Input tokens × Input multiplier + Output tokens × Output multiplier. The input/output multipliers follow the per-token rates published on each model's detail page on our website, and vary by model.
For the latest list of supported models and point multipliers, please refer to the "Supported Models and Multipliers" table on this page.
Coding Plan supports a variety of mainstream programming tools, such as Claude Code, Cursor, Cline (VSCode), Codex CLI, Kilo Code, Roo Code, and OpenCode.
Once your monthly plan limit is reached — whether the weekly cap or the monthly total — you can purchase a pay-as-you-go package, and any excess usage is deducted from it. If you have no pay-as-you-go package, usage is paused until your credits refresh — the weekly cap refreshes every Monday at 00:00 UTC.
Monthly plans use a dual limit — a monthly total plus a weekly cap. The weekly cap refreshes every Monday at 00:00 UTC, and any unused weekly allowance does not carry over to the next week; the monthly total stays available throughout the plan period and is cleared when the plan expires. Pay-as-you-go package points accumulate and remain usable throughout their validity period.
Coding Plan keys are restricted to models included in your subscription and cannot call models outside the plan, which avoids unexpected charges. When creating an API Key on the API Keys page, please choose the "Coding Plan" scope — it is managed separately from pay-as-you-go keys.
