Pricing · value first

Pick quota by the size of your tasks

Compare allowances, tool access and concurrency for your current workload.

Choose a plan

免费版

Free

Try the API and basic stats

CN¥0/ 30 days

1M equivalent tokens

  • Concurrency 2
  • 20 requests / min
  • SDK / curl
  • Agent runs inside your terminal
Features and limits

    入门版

    Personal projects and common terminals

    CN¥0.01/ 30 days

    20M equivalent tokens

    • Concurrency 10
    • 120 requests / min
    • SDK / curl
    • Agent runs inside your terminal
    Features and limits

      专业版

      Several projects and higher concurrency

      CN¥99/ 30 days

      100M equivalent tokens

      • Concurrency 50
      • 600 requests / min
      • SDK / curl · Cursor
      • Agent runs inside your terminal
      Features and limits
      • Project profile injection
      • Task merging

      Pro Plus

      Recommended

      The daily driver for team coding

      CN¥199/ 30 days

      250M equivalent tokens

      • Concurrency 20
      • 600 requests / min
      • SDK / curl · Cursor · Claude Code · Codex CLI
      • Agent runs inside your terminal
      Features and limits
      • Claude Code access
      • Codex CLI access
      • Project profile injection
      • Task merging
      • Playground thinking mode
      • Usage export
      • Difficulty-aware assistance
      • Burst concurrency: 40

      Regular concurrency is 20, below 专业版 at 50. Consider this alongside the allowance and included features.

      Pro Max

      Heavy agent runs and cloud execution

      CN¥599/ 30 days

      800M equivalent tokens

      • Concurrency 50
      • 1200 requests / min
      • SDK / curl · Cursor · Claude Code · Codex CLI
      • Cloud agents · workflows
      Features and limits
      • Claude Code access
      • Codex CLI access
      • Project profile injection
      • Task merging
      • Playground thinking mode
      • Usage export
      • Difficulty-aware assistance
      • Burst concurrency: 100
      • Priority queue

      Free plans roll over automatically. Paid plans follow your auto-renewal setting and require sufficient wallet balance to renew.

      A higher price does not mean every limit is higher. Compare feature access, allowance, regular concurrency and burst concurrency before upgrading.

      How a task uses your allowance.

      How it works

      You buy more than tokens: a more complete coding process

      Usage is measured in equivalent tokens. Supported clients, project memory and cloud execution vary by plan. Check the feature comparison before choosing an allowance.

      Understand the project, then generate

      Repository structure, team conventions and past decisions enter the current task; no re-explaining in every chat.

      Route models by task stage

      Search, analysis, implementation and verification take different paths; hard steps get the strong model, routine steps do not burn quota.

      Keep the terminal you know

      Codex CLI, Claude Code and common SDKs plug into the same project context, memory and call ledger.

      Start freeFree plan includes 1M equiv. tokens · no per-seat purchase
      TASK METERING EXAMPLE
      Example allowance calculation for a bug fix task15.6K equiv. tokens
      1. 01
        Search project
        Fast
        800
        done
      2. 02
        Analyse dependencies
        Fast
        1.4K
        done
      3. 03
        Generate fix
        Deep
        3.2K
        running
      4. 04
        Verify tests
        Fast
        600
        waiting
      Example cost
      —
      Amount shown once unit prices are published
      Fast · 2.8K1×
      Deep · 3.2K4×
      Sample data, not a real task record
      Beyond the quota: CN¥4.50 per million equiv. tokens from the wallet

      Pricing

      Capability comparison

      Compare the features actually included in each plan. Excluded features show their limitations.

      免费版

      • Automatic project profile and memory injection· Higher plan required
      • Per-call detail and cost
      • Anthropic Messages / Responses protocols· Higher plan required
      • Codex CLI · Claude Code· Higher plan required
      • Burst concurrency· Higher plan required
      • Cloud agents and workflows· Higher plan required
      • Priority queue· Higher plan required

      入门版

      • Automatic project profile and memory injection· Higher plan required
      • Per-call detail and cost
      • Anthropic Messages / Responses protocols· Higher plan required
      • Codex CLI · Claude Code· Higher plan required
      • Burst concurrency· Higher plan required
      • Cloud agents and workflows· Higher plan required
      • Priority queue· Higher plan required

      专业版

      • Automatic project profile and memory injection
      • Per-call detail and cost
      • Anthropic Messages / Responses protocols· Higher plan required
      • Codex CLI · Claude Code· Higher plan required
      • Burst concurrency· Higher plan required
      • Cloud agents and workflows· Higher plan required
      • Priority queue· Higher plan required

      Pro Plus

      • Automatic project profile and memory injection
      • Per-call detail and cost
      • Anthropic Messages / Responses protocols
      • Codex CLI · Claude Code
      • Burst concurrencyup to 40
      • Cloud agents and workflows· Higher plan required
      • Priority queue· Higher plan required

      Pro Max

      • Automatic project profile and memory injection
      • Per-call detail and cost
      • Anthropic Messages / Responses protocols
      • Codex CLI · Claude Code
      • Burst concurrencyup to 100
      • Cloud agents and workflows
      • Priority queue