Scale your swarm without scaling your bill
Predictable agent pricing. Every plan includes the full model matrix — OpenAI, Anthropic, Azure, Google, xAI, DeepSeek, Moonshot, Groq, Mistral, Cohere — with rate limits and quota controls designed for production teams. Metered overage at cost+20% when you burst past the included credit pool.
Starter
Evaluate the swarm
- 500 credits one-time
- Bring your own keys (BYOK)
- 1 concurrent project
- Community support
Pro
Solo pro · most popular
- $75 in credits every month
- All providers: OpenAI, Anthropic, Azure, Google, xAI, DeepSeek, Moonshot, Groq, Mistral, Cohere
- Model pool routing (round-robin, weighted, cost, fastest)
- Unlimited projects
- Priority queue
- Metered overage at cost+20%
Team
For multi-agent startups
- $250 in credits every month
- Everything in Pro
- 5 agents included, $30/mo per extra agent
- Shared model pools + admin controls
- Audit log + role-based usage limits
- Engine-side rate limits per agent
Enterprise
Procurement-grade
- SSO + SAML
- Dedicated Azure quota
- SLA + 24h support
- Invoice billing
- Custom rate limits + overage cap
- Onboarded BYOK for orgs
FAQ
What's a credit?
1 credit = $0.001 USD. Token billing: each model has a per-1K-token input/output rate applied against your balance. Included plan credits are renewed monthly. Overage is metered and charged at cost+20% on your next invoice.
Can I bring my own API keys?
Yes — BYOK is first-class on every plan. Pro and Team include all 11 providers. You only burn SwarmAgents credits on platform-managed models; your own-provider traffic is free of platform charges. Rate limits and pool routing still apply.
How does model pool routing work?
Each role (commander, coder, verifier, researcher, browser, summarizer) can either pin a single model or rotate through a pool. Pool strategies: round-robin, weighted random, least-cost, or fastest. Routing resolves on the engine, not the browser — no client-side key leaks.
What happens if I hit my cap?
Workloads pause — never surprise-bill you. Bump the cap in settings, or let the next billing cycle top you up. Plans can be upgraded, downgraded, or canceled at any time from the Customer Portal.
Model prices
Credits charged per 1K tokens. Prices marked 20% over Azure list as of 2026-08.
| Model | Input / 1K | Output / 1K | Azure list (in/out) |
|---|---|---|---|
| Llama-3.2-1B | 0.03 cr | 0.03 cr | $0.00002 / $0.00002 |
| GPT-5.4-Nano | 0.07 cr | 0.56 cr | $0.00005 / $0.00040 |
| Phi-4-mini | 0.11 cr | 0.21 cr | $0.00007 / $0.00015 |
| Gemma-3-12B | 0.18 cr | 0.53 cr | $0.00013 / $0.00038 |
| Phi-3.5-mini | 0.19 cr | 0.73 cr | $0.00013 / $0.00052 |
| DeepSeek-V4-Flash | 0.20 cr | 0.40 cr | $0.00014 / $0.00028 |
| Mistral-NeMo-12B | 0.21 cr | 0.21 cr | $0.00015 / $0.00015 |
| GPT-4o-mini | 0.21 cr | 0.84 cr | $0.00015 / $0.00060 |
| Qwen-3.5-9B | 0.28 cr | 0.84 cr | $0.00020 / $0.00060 |
| Llama-3.3-70B | 1.00 cr | 3.98 cr | $0.00071 / $0.00284 |
| DeepSeek-V4-Pro | 1.75 cr | 7.00 cr | $0.00125 / $0.00500 |
| DeepSeek-R1 | 1.89 cr | 7.56 cr | $0.00135 / $0.00540 |
| Qwen-3-Max | 2.24 cr | 8.96 cr | $0.00160 / $0.00640 |
| Llama-4-Maverick-405B | 3.78 cr | 3.78 cr | $0.00270 / $0.00270 |