Simple, transparent pricing
Start free with open models. Upgrade for higher limits, premium models, and enterprise features. No hidden fees.
Spark
Open models, 10K requests, 16K context. $1/month for new users — fully refundable.
Pro
Premium models, expanded context, cross-session memory, and Merge.dev Agent Handler.
Ultra
Maximum context, unlimited requests, premium models, and the highest availability.
Every plan. Side by side.
Every feature. Every limit. Every model. Exactly what you get on each plan.
| Feature | Spark | Pro | Ultra |
|---|---|---|---|
| Usage | |||
| Models | |||
| Performance | |||
| Analytics & support | |||
| Privacy & security | |||
| DeepSeek V4 Pro effective usagePermanent deal — credits go up to 4× further | ~$40 | ~$120 | ~$600 |
| MiniMax M3 effective usagePermanent deal — credits go up to 2.7× further | ~$27 | ~$80 | ~$400 |
| MiMo V2.5 effective usagePermanent deal — credits go up to 5× further | ~$50 | ~$150 | ~$750 |
| Typical request volumeApprox, depends on chosen models | ~10K | ~25K | ~110K |
| Roll-over unused credits | — | ✓ | ✓ |
| Premium models (Claude Opus 4.8, GPT‑5, Gemini Ultra 2, etc.) | — | ✓ | ✓ |
| taste-1 — learns and applies your coding style | ✓ | ✓ | ✓ |
| Agent Handler (Merge.dev integration) | — | ✓ | ✓ |
| Context window per request | 16K tokens | 128K tokens | 1M tokens |
| Concurrent requests | 3 | 10 | 50 |
| Auto top-up at API cost | — | — | ✓ |
| Per-request cost & token breakdown | — | ✓ | ✓ |
| Discord community support | ✓ | ✓ | ✓ |
| Email support | — | ✓ | ✓ |
| Priority support (4h response) | — | — | ✓ |
| Local-only taste storage | ✓ | ✓ | ✓ |
| SOC 2 compliance | — | — | ✓ |
| Cancel anytime — no lock-in | ✓ | ✓ | ✓ |
Need a side-by-side per-model breakdown? See full pricing & limits docs or open the usage calculator.
Enterprise
Tailored pricing and infrastructure for organizations that need custom credits, unlimited seats, dedicated support, and SLA guarantees.
Contact SalesFrequently asked questions
Spark is $1/month for new users — fully refundable. It is free for existing (grandfathered) users with 10K requests, 16K context, and $5 monthly credits. Upgrade to Spark Premium ($1/month) for 15K requests, 32K context, and $10 credits.
We offer location-based pricing for the Pro plan. Indian users pay ₹749/month (or ₹8,300/year) via UPI. International users pay $12/month (or $140/year) via card. All other plans are priced uniformly worldwide.
Yes. Your plan changes take effect immediately. Upgrades are prorated; downgrades credit your account for the remaining billing period.
Spark grants access to open-weight models (Llama, DeepSeek, Mistral, etc.). Pro adds premium models (Claude Opus, GPT-4o, Gemini Ultra) with expanded context. Ultra unlocks the full model catalog at maximum context.
Pro and Ultra plans include integration with Merge.dev's Agent Handler, giving your agents unified access to 100+ enterprise tools and APIs through a single interface.
Pro and Ultra come with a 7-day free trial. No credit card required to start. Cancel anytime during the trial — you won't be charged.
A request is a single call to a model (including streaming conversations, tool calls, and agent spawns). Token-based usage is tracked separately; see your dashboard for real-time limits.