Founding batch · flat-rate plans
$12–$99/month · ₹1,999/month in India. GLM-5.3 (755B), DeepSeek V4.1-Flash and Qwen3.8-27B on one OpenAI-compatible key — no per-token meter, no surprise bills. A GPU node only runs when its seats are prepaid.
0/45
seats confirmed — node warms at 45
$12
from, per month — every plan unlimited tokens
₹1,999
Interactive in India — UPI + GST invoice
12 mo
founding price lock · cancel anytime
Counter is live — farm and test accounts excluded. The flash node warms at 45 seats (prepaid ≥ 1.3× its $824/mo cost).
Every plan: unlimited tokens within fair use, the full model catalog, one OpenAI-compatible key. The tier sets your sustained speed lane.
Agents
BATCH-GATED$12/mo
Unlimited tokens · 20 tok/s lane
Background agents, evals, batch jobs, RAG pipelines.
InteractiveMost picked
BATCH-GATED$24/mo
₹1,999/mo in India (UPI + GST invoice)
Unlimited tokens · 60 tok/s lane
Chat products, coding assistants, prototyping.
Fast
BATCH-GATED$49/mo
Unlimited tokens · 120 tok/s lane
Production apps, live traffic, real-time voice.
Founding Flagship
LIVE NOW$99/mo
Unlimited tokens · 60 tok/s lane
Frontier-model products and hard reasoning workloads.
Need 100M–1B tokens/day of coding-agent burn? Token-Max plans →
Every plan serves the full live catalog — unlimited tokens within fair use. Pay-as-you-go rates shown for comparison.
| Model | Detail | Pay-as-you-go | On flat-rate plans |
|---|---|---|---|
| GLM-5.3 | 755B frontier · 202K context | $1.40 in / $4.40 out per 1M | Unlimited tokens* |
| DeepSeek V4.1-Flash | 1M context · 218 tok/s | $0.22 in / $0.66 out per 1M | Unlimited tokens* |
| Qwen3.8-27B Flash | 32K context | $0.40 in / $1.60 out per 1M | Unlimited tokens* |
*Within the 10M/day fair-use lane. Speed-lane models publish with benchmark receipts the day the flash node lights — we never list a model that 404s.
Takes 30 seconds. Confirmation lands in your inbox with your reservation ID.
Prefer pay-as-you-go? $5 free credits → · Building a startup? Credits program →
Every plan includes the full served catalog on one OpenAI-compatible key: GLM-5.3 (755B, 202K context), DeepSeek V4.1-Flash (1M context) and Qwen3.8-27B Flash. Your tier sets your sustained speed lane, not your model list. Founding Flagship adds the priority lane on GLM-5.3, which runs on dedicated 8-GPU hardware today.
Because no GPU node runs before its seats are prepaid. 45 confirmed seats is 1.3× the node's $824/month cost — the node warms the day the counter fills, and not a day earlier. If the batch never fills, every deposit is refunded in full.
It verifies you're a real human with a real card, counts your seat in the public counter, and refunds to your wallet with a $10 bonus when your plan activates. Farm and test accounts are excluded from the counter by pattern.
Up to 10M tokens per day your speed tier holds. Above the soft cap nothing extra bills — burst traffic shifts to the agents lane so interactive speeds stay honest. Sustained 100M+/day workloads get a dedicated-node offer instead of a throttle.
Interactive is ₹1,999/month via UPI with a GST invoice — the same plan on regional rails. Every tier is cancel-anytime monthly, and founding prices lock for 12 months from activation.
You get a confirmation email with your reservation ID. Create your account, verify your card ($1, refunded instantly, +$10 bonus), and your seat counts. The flagship lane activates immediately; flash tiers activate the day the batch fills.