Enterprise-grade inference for teams that can't compromise
Run production AI workloads on an OpenAI-compatible API with transparent token pricing — or reserve dedicated capacity engineered around your security, regional, and performance requirements.
Built by Alchemy, the company powering blockchain infrastructure for some of the world's largest companies—with more than 1 trillion transactions landed and over 500 million end users powered worldwide.
Two ways to run inference
Token APIs
Self-serve, usage-based access to deployed models.
- • OpenAI-compatible API — switch by changing one base URL
- • Pay per token with transparent input/output pricing
- • App-scoped API keys with rotation, expiry, and instant revocation
- • Spending and rate limits at organization and app level
- • Usage analytics, credits, and pay-as-you-go billing
Dedicated Deployments
Reserved capacity for enterprise workloads.
- • Capacity reserved for your organization — no noisy neighbors
- • Deployment options shaped to your compliance requirements
- • Regional placement to meet data-residency needs
- • Enterprise support with a named team
- • Performance targets agreed together in your contract
Deployed models, transparent pricing
Prices are per million tokens, billed separately for input and output. Development catalog shown for illustration — the live catalog is served from your deployment.
| Model | Context | Capabilities | Input / Mtok | Output / Mtok |
|---|---|---|---|---|
Alchemy Large 1 alchemy-large-1 | 200,000 tokens | chat, tool-use, json-mode | $3.00 | $15.00 |
Alchemy Small 1 alchemy-small-1 | 128,000 tokens | chat, json-mode | $0.25 | $1.25 |
Alchemy Vision 1 alchemy-vision-1 | 128,000 tokens | chat, vision | $1.00 | $5.00 |
OpenAI-compatible. Migrate in minutes.
Point your existing OpenAI SDK at Alchemy Inference and keep your code. One app key works across every available model via the model parameter, and keys stay scoped to a single app and organization with inference-only permissions.
from openai import OpenAI
client = OpenAI(
base_url="https://<your-deployment>/api/v1",
api_key=os.environ["AINF_API_KEY"], # never hardcode keys
)
response = client.chat.completions.create(
model="alchemy-large-1",
messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)Security by default
- • Keys generated server-side with 256-bit entropy, stored only as verifiers
- • Secrets shown once; instant revocation and rotation with migration windows
- • Authorization, expiry, and tenant isolation enforced server-side
- • TLS everywhere; key lifecycle fully audited
Operational reliability
- • Enforceable spending and rate limits at org and app level
- • Billing failures surface honestly — never as silent success
- • Idempotent billing operations; retries can't double-charge
- • Clear model availability status in the catalog
Built for performance
- • Streaming responses with cancellation support
- • Latency metrics alongside token usage in analytics
- • Dedicated capacity available when shared limits don't fit
Ready to build?
Create an organization, generate an app key, and make your first request in minutes — or talk to us about dedicated capacity.