gpt-oss-120b API — Fast, Private, Attested Inference
Run gpt-oss-120b through an OpenAI-compatible API: a Cerebras-served fast route, TEE-attested privacy tiers, and measured latency from real requests.
gpt-oss-120b, served fast on Cerebras and attested down to the image digest.
gpt-oss-120b is OpenAI's open-weight model, and the list of providers selling API access to it keeps growing. Comparison sites rank every host on price and speed. TrustedRouter competes on both, then adds the axis those tables leave out: proof of where your prompt runs and who can read it.
You get two routes. openai/gpt-oss-120b for standard serving, and cerebras/gpt-oss-120b, a fast route for latency-sensitive work. Both sit behind one OpenAI-compatible endpoint: point base_url at the gateway, keep your SDK, your model ids, and your code. Measured latency from real routed requests is published on the latency benchmarks page, so the speed claim comes with numbers you can inspect.
from openai import OpenAI
client = OpenAI(
base_url="https://api.trustedrouter.com/v1",
api_key="YOUR_TRUSTEDROUTER_KEY",
)
resp = client.chat.completions.create(
model="cerebras/gpt-oss-120b",
messages=[{"role": "user",
"content": "Summarize this filing."}],
)
print(resp.choices[0].message.content)A fast route with published numbers
The cerebras/gpt-oss-120b route exists for one reason: latency. When a provider goes down, automatic fallback rolls your request to another route. The data on the latency benchmarks page comes from real routed requests, so what you compare is what you get in production.
Attestation you can check live
The prompt path runs inside a hardware-attested confidential-compute gateway. The source is open, the image digest is published, and the attestation endpoint returns a JWT signed by the CPU vendor's root key, bound to your live TLS session. TrustedRouter does not store prompt or output content by default. Details at /security.
HIPAA-ready architecture, stated plainly
TrustedRouter publishes HIPAA readiness documentation, a BAA, a DPA, and a subprocessor list, and states plainly that readiness documentation is not an audit report. Your compliance team assesses a verifiable architecture, with Zero-Data-Retention and TEE privacy tiers available on this model.
Pick the route. Keep the fallback.
Pricing is per model and published at /pricing, a thin markup over provider list prices. Pay with prepaid credits, bring your own provider key, or use x402 pay-per-request with MetaMask wallet sign-in. When a provider degrades, requests roll over automatically, so a single host's bad day never becomes your outage.
The architecture holds up under load: a document-processing company scaled from zero to over a billion tokens per week through the attested gateway within three weeks of onboarding. Weigh gpt-oss-120b against every other route at /choose. Privacy you can check beats privacy you're promised.
Questions
What is the fastest way to run gpt-oss-120b through an API?
TrustedRouter serves cerebras/gpt-oss-120b, a fast route on Cerebras hardware, alongside the standard openai/gpt-oss-120b route. Rather than quoting a single headline number, TrustedRouter publishes measured latency from real routed requests on its provider latency benchmarks page, so you can compare routes on data before committing. Automatic provider fallback keeps requests flowing when a provider degrades.
How much does the gpt-oss-120b API cost on TrustedRouter?
Pricing is per model and published on the pricing page, with a thin markup over provider list prices. Prices drift, so check the live page rather than a cached table. You can pay with prepaid credits through Stripe, bring your own provider key, or use x402 pay-per-request crypto with MetaMask wallet sign-in, which works without a traditional account.
Can I use gpt-oss-120b for HIPAA-sensitive workloads?
TrustedRouter does not claim HIPAA certification. It offers a verifiable architecture plus readiness documentation your compliance team can assess: a hardware-attested TEE prompt path, no storage of prompt or output content by default, and HIPAA readiness, BAA, DPA, and subprocessor pages your reviewers can read directly before making their own determination.
Do I have to change my code to use TrustedRouter's gpt-oss-120b API?
No. The API is OpenAI-compatible. Change the base_url in your existing SDK to TrustedRouter's endpoint and keep your model ids and code; requests then route through the attested gateway. You can try gpt-oss-120b in the browser playground first, or use the model chooser to weigh it against other routes on the smart, cheap, and fast axes.