Agent Refill
Authorized agents can inspect capacity and activate a stored Reset Ticket without purchasing or entering billing.
PRICING
One AI platform for recurring capacity and prepaid tokens.
Choose recurring multi-model capacity for autonomous workloads, or prepaid Token Sachets for one-off AI usage—both with transparent routing and usage controls.
NO SUBSCRIPTIONPrefer prepaid, token-based access?Check out our Token Sachet for one-time AI usage. ↗Loading live plan catalog...
PREPAID OPTION
Token Sachet, when you want to buy once.
Prepaid AI usage with transparent model rates and no recurring subscription.
1 USD ≈ Rp—Loading live Token Sachet pricing...
PLAN COMPARISON
Compare ELTEX Agent Plan subscriptions.
| Detail | Go | Lite | Plus | Pro | Max |
|---|---|---|---|---|---|
| Price / month | $5 | $10 | $20 | $30 | $50 |
| Capacity window | 1,500 credits / 4-hour | 3,000 credits / 4-hour | 5,000 credits / 4-hour | 12,000 credits / 4-hour | 18,000 credits / 4-hour |
| Weekly capacity | 6,000 credits | 12,000 credits | 27,000 credits | 45,000 credits | 81,000 credits |
| Concurrent requests | 1 | 2 | 3 | 5 | 8 |
| Smart routing | Included | Included | Included | Included | Included |
| Force routing | Available | Available | Available | Available | Available |
| Agent Infrastructure | State + RunGuard + Agentic Wallet + authorized Refill | State + RunGuard + Agentic Wallet + authorized Refill | State + RunGuard + Agentic Wallet + authorized Refill | State + RunGuard + Agentic Wallet + authorized Refill | State + RunGuard + Agentic Wallet + authorized Refill |
| Model access | 8 configured models | 12 configured models | 15 configured models | 18 configured models | 18 configured models |
| Typical fit | For everyday AI tasks and light automation. | For consistent building, light coding, and analysis. | For larger workloads and deeper reasoning. | For production workflows and power users. | For teams and high-throughput AI operations. |
INCLUDED ON EVERY PAID PLAN
Agent Infrastructure for work that continues safely.
Agent State
Persist structured progress, decisions, checkpoints, private managed files, and artifact references across restarts.
Agent RunGuard
Enforce versioned cost, call, runtime, retry, failure, and repetition limits for agent runs.
Agentic Wallet
Grant one agent a dedicated key for policy-bounded USDC x402 payments on Base or Solana.
How capacity works
Plans use capacity refresh windows and weekly limits rather than exposing raw provider token pricing. A credit is a capacity unit, not a fixed quantity of tokens: use changes with the selected model, input and output tokens, context, tools, and route. Burst Pack capacity can be available when a plan window is exhausted.
COST REFERENCE
A reproducible DeepSeek-V4-Flash baseline.
This example uses DeepSeek's published direct-API rates for deepseek-v4-flash: US$0.14 per million cache-miss input tokens, US$0.0028 per million cache-hit input tokens, and US$0.28 per million output tokens. It is a reference calculation, not an ELTEX savings promise.
| Reference workload | Direct V4 Flash calculation | Direct API cost | What changes the result |
|---|---|---|---|
| 1,000,000 cache-miss input + 250,000 output | (1 × $0.14) + (0.25 × $0.28) | US$0.2100 | All input is billed at the cache-miss rate. |
| 1,000,000 cache-hit input + 250,000 output | (1 × $0.0028) + (0.25 × $0.28) | US$0.0728 | Input qualifies for DeepSeek cache-hit billing. |
Source: DeepSeek Models & Pricing, checked July 15, 2026. ELTEX will only publish savings percentages after they can be reproduced from a dated, route-level benchmark using equivalent workloads.