Indie
For buildersSide projects and small SaaS tools.
$59 / one time
$300Base Credits
80% below credit face value
Simple, prepaid pricingBase pricing for
Base pricing for
compatible AI APIs.
Base is the only active commercial tier. One-time payments, shared service credit and clear per-model token rates.
The active tier
Base Credits
For enabled Claude, GPT and Grok workloads.
ClaudeGPTGrok
Starter
A small start for your next idea.
$75Base Credits
$19one time
Pro
Production apps finding their stride.
$149 / one time
$1,000Base Credits
85% below credit face value
Scale
Larger workloads, teams and agencies.
$399 / one time
$3,000Base Credits
87% below credit face value
Base is active.
Customer credit and provider consumption use separate units. Primary routes use 1.0x customer pricing; an eligible fallback uses 1.5x customer pricing.
What you get per model
Published Base token rates
Rates are USD per 1,000,000 tokens. Input, output, cache reads and cache writes are separate components.
Enabled Base collection
Charged to your Base balance · USD per million tokens
Scroll across to compare all rates
| Model | Context / billing | Input / M | Cache read / M | Cache write / M | Output / M |
|---|---|---|---|---|---|
| Anthropic Claude | |||||
| claude-opus-5 | Flat rate | $5.00 | $0.50 | $10.00 | $25.00 |
| claude-opus-5-5 | Flat rate | $4.00 | $0.20 | $5.00 | $20.00 |
| claude-sonnet-5 | Flat rate | $2.00 | $0.20 | $4.00 | $10.00 |
| OpenAI GPT | |||||
| gpt-6-astra | ≤ 272K context | $10.00 | $1.00 | $12.50 | $50.00 |
| gpt-6-astra | > 272K context | $20.00 | $2.00 | $25.00 | $75.00 |
| gpt-5.6-terra | Flat rate | $2.00 | $0.20 | $0.20 | $12.00 |
| gpt-6.1-sol | Flat rate | $2.00 | $0.10 | $2.50 | $10.00 |
| gpt-5.5 | ≤ 272K context | $5.00 | $0.50 | $0.50 | $30.00 |
| gpt-5.5 | > 272K context | $10.00 | $1.00 | $1.00 | $60.00 |
| xAI Grok | |||||
| grok-4.5 | Flat rate | $2.00 | $0.50 | $0.50 | $6.00 |
| grok-4.6 | < 200K context | $2.00 | $0.50 | $0.50 | $6.00 |
| grok-4.6 | ≥ 200K context | $4.00 | $1.00 | $1.00 | $12.00 |
| DeepSeek | |||||
| deepseek-v4.1-flash | Peak rate | $2 | $0.04 | $0 | $8 |
| deepseek-v4.1-flash | Off-peak rate | $1 | $0.02 | $0 | $4 |
| Google Gemini | |||||
| gemini-3.5-flash | Flat rate | $0.375 | $0.0375 | $0 | $2.25 |
| gemini-3.1-pro | Flat rate | $0.5 | $0.05 | $0.5 | $3 |
| Z.ai GLM | |||||
| glm-5.3 | Flat rate | $1.4 | $0.26 | $0 | $4.4 |
| glm-5.3-flash | Flat rate | $0.15 | $0.03 | $0 | $0.5 |
| Moonshot Kimi | |||||
| kimi-k3 | Flat rate | $3 | $0.3 | $0 | $15 |
Create with GPT Image
Image generation and editing
One Base rate per successful generation or accepted edit.
gpt-image-2.5-flare
Fast everyday generation and edits.
$0.30per request
gpt-image-2.5-sunburst
Detailed visual generation and edits.
$0.30per request
Use a GPT key. See image API examples →
Fallbacks. OpenAI, Claude and Grok are fallback provider families. A fallback mapping is admitted only with a valid pool, group, compatible endpoint, known tariff and limits, and sufficient customer and native reserves. The current active fallback mapping is OpenAI-compatible GPT gpt-6.1-sol; Claude and Grok fallback mappings remain disabled until their mapping evidence is complete.
DeepSeek dynamic rates. Base routes deepseek-v4.1-flash to the same upstream model ID. Rates per million tokens are $2 input, $8 output and $0.04 cache read at peak; $1 input, $4 output and $0.02 cache read off-peak. Peak hours are Monday through Friday, 01:00–04:00 and 06:00–10:00 UTC, excluding Chinese public holidays. The gateway fixes the period when the request arrives. DeepSeek Pro remains temporarily unavailable.
DeepSeek rates updated on . Context limits and availability depend on confirmed route mappings. Choose a listed model ID; unsupported or pending identifiers are rejected safely.
Service credit and availability
You are purchasing prepaid use of Zrelay’s managed AI API gateway. Customer service credit and native provider consumption are independent accounting units. A primary request is charged at 1.0x. An eligible fallback is charged at 1.5x to the customer while its native provider cost remains separate.
Only enabled models are available. See the model IDs and rates above.