Pricing Explained
How NexGate credits, token pricing, credit packs, reservations, billing errors, and usage exports work.
How credits work
NexGate uses prepaid credits. You buy credits first, then each API call deducts credits based on token usage and the configured model price.
Note
Credits are USD-denominated. Every credit equals $1 USD of API usage across any supported model.
Buy credits
Purchase credit packs starting at $25.
Use credits
Chat requests deduct credits based on actual token usage.
Keep control
Requests stop when credits or spend safety limits are reached.
Credit packs
| Pack | Price | Credits | Best for |
|---|---|---|---|
| Builder | $25 | $30.00 | Side projects and prototypes |
| Pro | $50 | $60.00 | Production apps |
| Team | $100 | $120.00 | High-volume applications |
| Scale | $200 | $250.00 | Growing products |
Tip
The price you pay is the amount charged; the credits column is what lands in your account. You pay $25 and get $30.00 in credits. Credits never expire.
Note
Eligible new users receive $5 in free credits after sign-up — enough to make hundreds of test requests before buying a pack.
Buy credits
Open the live pricing page.
How token pricing works
Every API call is priced from:
- The resolved model.
- Input tokens.
- Output tokens.
- The NexGate price configured at the time of use.
Example calculation
A gpt-5.5 request with 500 input tokens and 200 output tokens costs:
Input: 500 / 1,000,000 x $5.00 = $0.002500
Output: 200 / 1,000,000 x $30.00 = $0.006000
Total: $0.008500Your dashboard and CSV export show the exact logged cost for each request.
Credit reservations
Before NexGate calls a provider, it reserves credits using estimated input tokens and the maximum possible output. After the response completes, NexGate charges actual usage and releases unused reserved credits.
reserve estimated maximum cost -> provider response -> charge actual usage -> release remainderreserve estimated maximum cost -> provider error -> release reservation -> log errorreserve estimated maximum cost -> stream disconnects before usage -> release reservation -> log client_disconnectedTip
Set max_tokens or max_completion_tokens to reduce the maximum reservation for a request.
Token pricing by model
All prices are per 1 million tokens.
| Model | Input / 1M | Output / 1M | Use case |
|---|---|---|---|
| GPT-5.5 | $5.00 | $30.00 | Latest deployed GPT-5.5 flagship |
| GPT-5 | $1.25 | $10.00 | GPT-5 reasoning and coding |
| GPT-5.4 | $2.50 | $15.00 | Updated GPT-5.4 model |
| GPT-4o | $2.50 | $10.00 | Multimodal flagship |
| GPT-4.1 | $2.00 | $8.00 | Long-context production reasoning |
| o4 Mini | $1.10 | $4.40 | Fast reasoning planner |
| GPT-5.4 Mini | $0.40 | $1.60 | Fast GPT-5.4 workhorse |
| GPT-4.1 Mini | $0.40 | $1.60 | Reliable workhorse |
| GPT-5.4 Nano | $0.10 | $0.40 | Ultra-low-cost GPT-5.4 tasks |
| GPT-4.1 Nano | $0.10 | $0.40 | Classification and routing |
| DeepSeek V4 Pro | $1.93 | $3.83 | Frontier DeepSeek V4 with enhanced reasoning |
| DeepSeek V3.2 Speciale | $0.40 | $0.80 | Higher-capability DeepSeek chat |
| DeepSeek V3.2 | $0.28 | $0.42 | Efficient coding and chat |
| DeepSeek V4 Flash | $0.14 | $0.28 | Fast low-cost chat and coding |
| Kimi K2.6 | $0.95 | $4.00 | Long-context agent chat |
| Kimi K2.5 | $0.60 | $2.50 | Long-context multilingual chat |
| Grok 4.3 | $1.25 | $2.50 | Latest Grok 4.3 frontier model |
| Grok 4 20 Reasoning | $1.25 | $2.50 | Frontier reasoning and analysis |
| Grok 4.1 Fast Reasoning | $0.20 | $0.50 | Fast reasoning workloads |
| Llama 3.3 70B Instruct | $0.13 | $0.40 | Meta Llama 3.3 chat |
| Llama 4 Maverick | $0.18 | $0.70 | Open MoE model |
| Llama 4 Scout | $0.10 | $0.35 | Low-cost open chat |
Warning
Model prices can change when upstream provider costs change. Credits stay USD-denominated and requests are charged at the NexGate price configured at the time of use.
No automatic billing
NexGate does not automatically charge your card when credits run low. When credits are insufficient:
- Chat completion requests return
402 insufficient_credits. - Low-balance alerts can notify you if you set a threshold.
- You manually buy more credits when ready.
Spend safety limits
NexGate enforces an hourly spend safety limit on your account. If the estimated request would exceed that limit, chat completions return 429 rate_limit_error.
Note
You can set a finite hourly limit or use -1 for unlimited through the dashboard account settings flow.
You can also set optional per-key hourly limits — a dollars-per-hour spend ceiling and a requests-per-hour ceiling — on each API key. A request is rejected with 429 rate_limit_error if it would exceed either the account limit or any limit on the key it authenticated with.
Purchase capacity
NexGate sells credits against a configurable platform-wide capacity. When capacity is temporarily sold out, the pricing and top-up pages show a "capacity full" state and disable new purchases. Existing credits remain fully usable, and new purchases reopen once capacity is available again.
Refunds and expiration
Credits do not expire. Purchased credits are non-refundable, except where support determines a technical billing issue caused incorrect charging.
Warning
Do not buy credits expecting automatic refunds for unused balance.
Promo credits
Eligible new users may receive launch promo credits after sign-up verification. Promo availability and amount are controlled by runtime settings and anti-abuse checks.
Usage tracking
Your dashboard shows:
- Real-time balance
- Per-request cost
- Prompt and completion token counts
- Resolved model and requested alias when applicable
- Latency
- Status
- CSV export for the latest 1,000 usage rows
date,model,requested_model,prompt_tokens,completion_tokens,cost_usd,latency_ms,statusCost optimization
Billing error states
| Status | Error | Meaning |
|---|---|---|
402 | insufficient_credits | Available balance cannot cover the estimated request |
429 | rate_limit_error | Hourly spend safety limit has been reached |
503 | This pack is not yet configured. Please contact support. | Selected credit pack is temporarily unavailable |
500 | Server misconfigured | A configuration error prevents checkout from completing |