AI models behind one key, with a cap you set
Call AI models through a chat completions API. Each request is reserved on your prepaid balance and settled at actual usage, with caps per key.
Prepaid, never billed after
Galon AI is a chat completions and embeddings API for hosted models: each request is reserved on your prepaid balance before the call and settled at the usage actually received, behind the same key as the rest of Galon Cloud.
Updated . Prices and limits are read from the served price list and the platform code when this page is built.
| Plan or model | Included | Price | Per unit |
|---|---|---|---|
| Meta-Llama-3_3-70B-Instruct, input | Pay as you use | $1.1976 per million input tokens | — |
| Meta-Llama-3_3-70B-Instruct, output | Pay as you use | $1.1976 per million output tokens | — |
| gpt-oss-120b, input | Pay as you use | $0.1485 per million input tokens | — |
| gpt-oss-120b, output | Pay as you use | $0.7425 per million output tokens | — |
| gpt-oss-20b, input | Pay as you use | $0.0742 per million input tokens | — |
| gpt-oss-20b, output | Pay as you use | $0.2784 per million output tokens | — |
| Qwen2.5-VL-72B-Instruct, input | Pay as you use | $1.6266 per million input tokens | — |
| Qwen2.5-VL-72B-Instruct, output | Pay as you use | $1.6266 per million output tokens | — |
| Qwen3.5-397B-A17B, input | Pay as you use | $1.0725 per million input tokens | — |
| Qwen3.5-397B-A17B, output | Pay as you use | $6.435 per million output tokens | — |
| Qwen3.5-9B, input | Pay as you use | $0.1787 per million input tokens | — |
| Qwen3.5-9B, output | Pay as you use | $0.2681 per million output tokens | — |
| Qwen3.6-27B, input | Pay as you use | $0.715 per million input tokens | — |
| Qwen3.6-27B, output | Pay as you use | $4.8262 per million output tokens | — |
| Qwen3.8-27B, input | Pay as you use | $0.7425 per million input tokens | — |
| Qwen3.8-27B, output | Pay as you use | $4.8262 per million output tokens | — |
| bge-m3, input | Pay as you use | $0.0186 per million input tokens | — |
| Qwen3-Embedding-8B, input | Pay as you use | $0.1856 per million input tokens | — |
| Limit | Value |
|---|---|
| Requests per project and day (default quota) | 10,000 |
| Requests per project and month (default quota) | 100,000 |
| Requests per key and minute (sliding window) | 60 |
| Tokens per key and minute (input bound + max output, sliding window) | 200,000 |
| Input per request (messages, default) | 64 KiB |
| Output tokens per request (Galon aliases) | 8,192 |
| Messages per request | 256 |
| Tools per request | 128 |
| Images per request | 8 |
| Size of one image | 5 MiB |
| Request body | 2 MiB |
AI with a budget
- Chat completions and embeddings One key, one API, the same scopes and limits as the rest of Galon
- Models listed, not guessed The models endpoint returns what is served: real name, publisher, licence, context length and price
- Reserved before the call Each request is reserved on the prepaid balance, then settled at the usage received
- Streaming For models that publish it; an interrupted stream is billed for what was received only
- A cap per key Lower than the project limit if you want
Three steps
- Create your accountFree, no card
- Copy your MCP configPaste it into your AI assistant
- Pick a modelList them, call one
- Works through MCP, CLI or API
- Prepaid, never billed after
- Your cap, alert at 75%
curl -X POST https://api.galon.ai/v1/ai/chat/completions \ -H "Authorization: Bearer gln_test_••••" \ -H "Content-Type: application/json" \ -d '{"model": "{model}", "messages": [{"role": "user", "content": "Write a two-line menu description for an espresso."}]}' # the models served, with their prices curl https://api.galon.ai/v1/ai/models \ -H "Authorization: Bearer gln_test_••••"
# your coding agent, Galon's MCP server connected > Which models can I call, and what do they cost per million tokens? # the agent calls list_ai_models, get_pricing
curl -X POST https://api.galon.ai/v1/ai/embeddings \
-H "Authorization: Bearer gln_test_••••" \
-H "Content-Type: application/json" \
-d '{"model": "{model}", "input": ["espresso", "flat white", "cold brew"]}'Prepaid, pay as you use
Meta-Llama-3_3-70B-Instruct
$1.1976 per million input tokens
Meta-Llama-3_3-70B-Instruct
$1.1976 per million output tokens
gpt-oss-120b
$0.1485 per million input tokens
gpt-oss-120b
$0.7425 per million output tokens
gpt-oss-20b
$0.0742 per million input tokens
gpt-oss-20b
$0.2784 per million output tokens
Qwen2.5-VL-72B-Instruct
$1.6266 per million input tokens
Qwen2.5-VL-72B-Instruct
$1.6266 per million output tokens
Qwen3.5-397B-A17B
$1.0725 per million input tokens
Qwen3.5-397B-A17B
$6.435 per million output tokens
Qwen3.5-9B
$0.1787 per million input tokens
Qwen3.5-9B
$0.2681 per million output tokens
Qwen3.6-27B
$0.715 per million input tokens
Qwen3.6-27B
$4.8262 per million output tokens
Qwen3.8-27B
$0.7425 per million input tokens
Qwen3.8-27B
$4.8262 per million output tokens
bge-m3
$0.0186 per million input tokens
Qwen3-Embedding-8B
$0.1856 per million input tokens
- No card to start
- At $0, paid actions pause and your data is kept 30 days
- Every cost reserved before the action runs
- 1% of our revenue funds carbon removal through Stripe Climate.
Embeddings from $0.0186 per million tokens
Prices are paid from your prepaid balance. Top-ups are sold through our payment partner, which issues your receipt; domain names are sold separately.
Straight answers
- How much does Galon AI cost?
- Prices are per million tokens and differ by model. Chat models start at $0.0742 per million input tokens and $0.2784 per million output tokens; embeddings start at $0.0186 per million input tokens. The table on this page lists every model served at a fixed price.
- How is usage billed?
- Each request is reserved on your prepaid balance before the call and settled at the usage actually received. A request the balance cannot cover is refused with a clear error.
- Which models can I call?
- The ones listed by
GET /v1/ai/models(or the MCP toollist_ai_models) for your project: 10 models are served at a fixed price today. Use the returned id; never guess one. - Does Galon AI support streaming?
- Yes, for models that publish it. An interrupted stream is billed for what was received only. Chat completions and embeddings use the same key, scopes and limits as the rest of Galon.
- Does Galon cap my spending?
- Galon sets no hard spending cap. You can set your own monthly cap for the project, and each key can carry a lower cap of its own.
- What are the request limits?
- Up to 60 requests per key and minute, 256 messages and 8,192 output tokens per request, and 8 images per request.
- MailTransactional email your agent can send by itself
- Mail ProProfessional mailboxes on your own domain
- DeployFrom a commit to a live address, driven by your agent
- StorageFiles for your app, uploaded straight to storage
- DomainsDomain names your agent can check and quote
- BackendA PostgreSQL database with migrations and backups
- FunctionsSmall JavaScript functions, isolated and bounded
- SearchWeb pages as clean Markdown for your app and your agent
- GeoMaps data for your app, behind one key
- ComputeIsolated compute for your workloads