Complex reasoning and large codebases
For cross-file understanding, complex refactors, deep code review, and long-context development.
- Opus, Sonnet, and Haiku tiers
- Native Messages and Count Tokens
AceDataCloud Coding Plan
One plan unlocks the complete Claude, GPT/Codex, GLM, Kimi, and DeepSeek coding fleet
ACE DATA CLOUD · MULTI-MODEL CODING
One metered balance. Every coding model.
Use one token across Claude Code, Codex, WorkBuddy, and compatible clients. Switch among Claude, GPT, GLM, Kimi, and DeepSeek by task. Input, output, and cache usage are billed separately, with per-call details.
Claude CodeGPT · CodexGLMKimiDeepSeekWorkBuddy & IDEs
80+currently callable models
1token and usage bill
4 TYPESinput, output, and cache detail
ALLequal access on Plus, Standard, Pro, and Max
The advantage is not betting on one ecosystem
Use efficient models for everyday work, then switch to premium models for architecture, complex refactors, and long context—without changing clients or bills.
Tool use and agent workflows
Responses and Chat Completions compatibility powers Codex CLI and OpenAI SDKs.
- Flagship and efficient GPT models
- Codex client-compatible entry point
Chinese engineering and daily coding
Combines Chinese understanding, code generation, and tool use for frequent engineering work.
- GLM 4.x and 5.x families
- Control cost per task
Long context and repository understanding
For long documents, large code contexts, planning, and multi-stage implementation.
- K2 and K3 model families
- Thinking and general models
Reasoning, algorithms, and frequent work
More choices for algorithm analysis, code reasoning, and cost-sensitive daily calls.
- R1, V3, and V4 families
- Choose speed or depth per task
Every plan unlocks every model
Plus, Standard, Pro, and Max unlock the same model fleet. They differ only in included Credits.
- One application and one token
- Renew by choice; no general-balance fallback
Compare membership sticker prices without confusing allowances
Pricing snapshot: 2026-08-11, USD converted at $1 ≈ ¥6.759674 before tax. Dynamic membership allowances, official API tokens, and Ace Credits are not equivalent units.
| Option | Monthly/30-day price | vs Ace sticker price | Best for | Main boundary |
|---|---|---|---|---|
| Claude Pro | about ¥135 ($20) | Ace Plus is ¥99, about 27% lower | Light Claude-only use | Dynamic allowance; capacity cannot be compared with 200 Credits |
| Claude Max 5x / 20x | about ¥676 / ¥1,352 | Ace Pro / Max are roughly the same | Heavy Claude-only interaction | Five-hour and weekly limits; no fixed token grant |
| Kimi Moderato / Allegro | ¥99 / ¥699 | Same sticker prices as Ace Plus / Pro | Kimi Code and consumer products | Shared monthly allowance plus five-hour/weekly controls |
| MiniMax Plus / Max / Ultra | ¥49 / ¥119 / ¥469 | Below all three Ace sticker prices | MiniMax models and multimodal users | Fixed five-hour and weekly windows |
| Cursor Pro / Pro+ / Ultra | about ¥135 / ¥406 / ¥1,352 | Ace Pro exceeds Pro+ and approaches Ultra | Cursor IDE and built-in agents | Includes $20 / $70 / $400 of other-model usage |
| AceDataCloud Coding | ¥99 / ¥699 / ¥1,399 | About 70%–89% lower for like-for-like official API tokens | Claude/GPT/Kimi/GLM/DeepSeek, automation, and per-call billing | No membership-style five-hour/weekly reset; API concurrency and rate limits still apply |
The defensible advantages are: Plus is about 27% below Claude Pro's sticker price, while like-for-like model tokens are about 70%–89% below official API pricing. Pro/Max are not cheap in the personal-subscription market; their value must come from multi-model access, transparent metering, and larger API budgets. Claude-only heavy users may get better value from Claude Max, while Kimi-only users may prefer a Kimi membership. Sources: Claude Code costs · MiniMax Token Plan · Cursor pricing · Kimi K3 pricing.
Community reports on Claude Max
One Max 5x user parsed 139 local JSONL files and measured about $2,894 in official API-equivalent usage over 16 days. One Max 20x user's local tracker recorded about $1,428 in API-equivalent usage during the month. These are heavy-user cases, not official allowances or promises for typical users. Max 5x source · Max 20x source
Transparent metering
Standard input, output, cache writes, and cache reads are recorded separately; repeated context costs less when it hits the cache.
Cross-model continuity
Do not put every task on one model. Use different models for daily and critical work, scripts, and automated workflows.
How much lower than official API pricing?
Claude Opus 5 is about 85%–89% lower; Kimi K3 is about 70%–79% lower. This compares the same model and token type, never a consumer-membership multiplier with Credits.
| Model, plan, and unit | Standard input | Output | Cache read | 5-minute cache write | vs official API |
|---|---|---|---|---|---|
| Claude Opus 5 official API · CNY/1M Tokens | ¥33.80 | ¥168.99 | ¥3.38 | ¥42.25 | — |
| Ace Plus · CNY/1M Tokens | ¥5.20 | ¥26.02 | ¥0.52 | ¥6.51 | about 85% lower |
| Ace Standard · CNY/1M Tokens | ¥4.91 | ¥24.56 | ¥0.49 | ¥6.14 | about 85% lower |
| Ace Pro · CNY/1M Tokens | ¥4.90 | ¥24.49 | ¥0.49 | ¥6.12 | about 86% lower |
| Ace Max · CNY/1M Tokens | ¥3.68 | ¥18.38 | ¥0.37 | ¥4.60 | about 89% lower |
| Kimi K3 official API · CNY/1M Tokens | ¥20.00 | ¥100.00 | ¥2.00 | — | — |
| Ace Plus · CNY/1M Tokens | ¥5.94 | ¥29.70 | ¥0.59 | — | about 70% lower |
| Ace Standard · CNY/1M Tokens | ¥5.61 | ¥28.03 | ¥0.56 | — | about 72% lower |
| Ace Pro · CNY/1M Tokens | ¥5.59 | ¥27.96 | ¥0.56 | — | about 72% lower |
| Ace Max · CNY/1M Tokens | ¥4.20 | ¥20.99 | ¥0.42 | — | about 79% lower |
Pricing snapshot: 2026-08-10. Claude's official USD prices are converted at $1 ≈ ¥6.759674 before tax; percentages use like-for-like token prices in the table. Actual totals depend on input/output mix, cache hits, tools, and price changes. This compares official APIs and does not mean Ace is cheaper than Claude Max or a Kimi membership.
What do Credits actually mean?
Use an industry active-day cost benchmark instead of pretending API logs reveal “hours per day.” Anthropic's official Claude Code cost guide reports an average of about $13 per developer per active day, with 90% of users below $30 per active day.
| Plan | Allowance | Like-for-like official API token value | At the $13 average active day | At a $30 high-intensity active day | Coverage of repeat-active sample |
|---|---|---|---|---|---|
| Coding Plus | 200 Credits | about $95 | about 7 active days | about 3 active days | 56% |
| Coding Standard | 640 Credits | about $304 | about 23 active days | about 10 active days | Everyday full-time tier |
| Coding Pro | 1,500 Credits | about $713 | about 55 active days | about 24 active days | 80% |
| Coding Max | 4,000 Credits | about $1,903 | about 146 active days | about 63 active days | 90% |
An active day is a workday with Claude Code API usage, not a fixed number of hours, messages, or tasks. The conversion uses the current Opus 5 ratio across all four token buckets; actual spend changes with model, context, cache, output, and tools. Platform coverage uses an anonymized 28-day sample of 41 repeat-active users. It is directional because the sample is small.
Four Credit allowances. The same complete model access.
Each plan is a one-time purchase. Credits remain valid for 30 days and do not auto-renew.
Coding Plus
200 credits
¥99 · Occasional personal development
- ✓ About 7 industry-average active days
- ✓ About $95 in like-for-like official API token value
- ✓ Short projects, tests, and occasional use
Recommended
Coding Standard
640 credits
¥299 · Everyday full-time development
- ✓ About 23 industry-average active days
- ✓ About $304 in like-for-like official API token value
- ✓ Sustained coding, repositories, and review over a work month
Coding Pro
1,500 credits
¥699 · Full-time high-intensity development
- ✓ About 24 industry high-intensity active days
- ✓ About $713 in like-for-like official API token value
- ✓ Sustained agents and frequent Opus over a work month
Coding Max
4,000 credits
¥1,399 · Fleet and team workload
- ✓ About 63 industry high-intensity active days
- ✓ About $1,903 in like-for-like official API token value
- ✓ Usually excessive for one person; built for agents and automation
Do not choose Pro unless you make sustained high-intensity calls on workdays. Do not choose Max unless you run multiple agents, a team workload, or an automated Fleet. These are reproducible budget references, not usage guarantees.
Keep your tools. Change two environment variables.
Claude Code uses native Messages; Codex, WorkBuddy, and compatible clients use OpenAI-compatible APIs.
Claude Code
export ANTHROPIC_BASE_URL="https://api.acedata.cloud" export ANTHROPIC_AUTH_TOKEN="YOUR_API_TOKEN" claude
Codex CLI · WorkBuddy · OpenAI SDK
export OPENAI_BASE_URL="https://api.acedata.cloud/v1" export OPENAI_API_KEY="YOUR_API_TOKEN" codex
Frequently asked questions
Is Coding Plan always cheaper than Claude Max or a Kimi membership?
No universal comparison is valid. Consumer memberships provide dynamic allowances; Ace Coding deducts Credits by token type. Claude-only users who consistently use the allowance may get better value from Claude Max. Ace Coding better fits multiple models, automation, or transparent per-call billing.
Do model permissions differ by plan?
No. Coding Plus, Coding Standard, Coding Pro, and Coding Max unlock the same complete model fleet. Plans differ only in included Credits.
Are the usage profiles a guarantee?
No. Consumption changes with model, context length, cache hits, output length, tools, and parallel tasks. Profiles only guide plan selection and do not promise fixed usage.
Will it automatically consume my general balance?
No. Coding applications disable general-balance fallback. You decide whether to purchase again.
Is this an official Claude, ChatGPT, or Kimi membership?
No. Coding Plan is a multi-model API plan from AceDataCloud. It is compatible with the clients and protocols but does not include or resell consumer subscription accounts.
Choose an official membership for one model. Choose Ace Coding for the full fleet.
One token connects multiple coding ecosystems with transparent, token-based billing.
Claude Messages Count Tokens API
POSThttps://api2.yujuntc.fun/v1/messages/count_tokens
POSThttps://api2.yujuntc.fun/claude/messages/count_tokens
Claude Count Tokens API calculates the number of input tokens.
Claude Messages API
POSThttps://api2.yujuntc.fun/v1/messages
POSThttps://api2.yujuntc.fun/claude/messages
Claude Messages API, and the official input and output are completely consistent.
OpenAI Chat Completion API
POSThttps://api2.yujuntc.fun/openai/chat/completions
POSThttps://api2.yujuntc.fun/v1/chat/completions
OpenAI Chat Completion API, compatible with the official API format.
Claude Chat Completion API
POSThttps://api2.yujuntc.fun/v1/chat/completions
POSThttps://api2.yujuntc.fun/claude/chat/completions
Claude Chat Completion API, compatible with the official API format.
DeepSeek Chat Completion API
POSThttps://api2.yujuntc.fun/deepseek/chat/completions
POSThttps://api2.yujuntc.fun/v1/chat/completions
DeepSeek Chat Completion API, compatible with the official API format.
GLM Chat Completion API
POSThttps://api2.yujuntc.fun/glm/chat/completions
POSThttps://api2.yujuntc.fun/v1/chat/completions
Zhipu GLM chat completion API, compatible with the OpenAI Chat Completions format. Supports models including glm-3-turbo, glm-4.5-air, glm-4.6, glm-4.7, and glm-5.1, with streaming, function calling, multi-turn conversations, and system prompts.
Gemini Chat Completion API
POSThttps://api2.yujuntc.fun/gemini/chat/completions
POSThttps://api2.yujuntc.fun/v1/chat/completions
Gemini Chat Completion API, compatible with the official API format.
Grok Chat Completion API
POSThttps://api2.yujuntc.fun/grok/chat/completions
POSThttps://api2.yujuntc.fun/v1/chat/completions
Grok Chat Completion API, compatible with the official API format.
Kimi Chat Completion API
POSThttps://api2.yujuntc.fun/kimi/chat/completions
POSThttps://api2.yujuntc.fun/v1/chat/completions
Kimi Chat Completion API, compatible with the official interface format.
OpenAI Responses API
POSThttps://api2.yujuntc.fun/openai/responses
POSThttps://api2.yujuntc.fun/v1/responses
OpenAI Responses API, compatible with the official API format.
Monthly Plans
One-time purchase. Credits remain valid for 30 days and do not auto-renew. Every tier unlocks the same models and clients.
One-time purchase · Credits valid for 30 days from purchase · No automatic renewal · All plans unlock the same model and client access
How to understand Credit usage
Credits are deducted from actual usage across standard input, output, cache writes, and cache reads. Consumption varies substantially with model, context length, cache hits, tools, generated output, and parallel tasks. Workload profiles guide selection and do not guarantee hours, days, messages, or tasks.
Introduction
APIs
Pricing