Managed inference API
Cerebras Inference
Managed inference built on Cerebras wafer-scale hardware. Published throughput figures depend on the model, context, precision and serving configuration.
Product facts checked October 2, 2026Editorial scores last reviewed September 3, 2026
Some outbound links use a first-party redirect hop for click counting. Commission is only claimed when a partner programme is active for that specific link — most vendor hops here are not paid placements. Affiliate disclosure.
Marketplace referrals (for example RunPod or Vast) may return credits or kickbacks when a programme is active — that is a material connection even when it is not a cash CPA.
Best for
Supported shared models or enterprise dedicated deployments where measured latency/throughput justify the serving path
Watch out
Confirm the served model and access route. Cerebras’s August 18, 2026 CS-4 announcement reports three WSE-3T wafers, up to 2x CS-3 speed and up to 10x throughput per watt; these are vendor-reported, workload-dependent figures. Its August 13 OpenAI Ultrafast announcement reports up to 750 output tokens/sec for GPT-5.6 Sol, initially in limited preview through the OpenAI API. The September 1 Finland announcement describes phased construction toward 165 MW.
| Billing | Shared Free Trial/Pay as You Go per-model tokens; enterprise Dedicated Inference reserves capacity under organization-specific pricing. |
|---|---|
| GPU choice | Cerebras wafer-scale serving infrastructure; hardware generation depends on the service |
| Cold start | Shared API and reserved dedicated deployments have different capacity behavior; no universal first-request latency is verified here. |
| Model access | Current Shared Inference docs list gpt-oss-120b and qwen-3.8-27b. Enterprise Dedicated Inference supports additional families and custom/fine-tuned weights matching a supported architecture, with separate endpoint IDs and capacity. |
Sources for the dated product and announcement claims
- Current product modes, billing and deployment documentation — source checked October 2, 2026
- Current product modes, billing and deployment documentation — source checked October 2, 2026
- Current product modes, billing and deployment documentation — source checked October 2, 2026
- CS-4 announcement and measurement caveats — source checked October 2, 2026
- OpenAI Ultrafast limited-preview announcement — source checked October 2, 2026
- Finland phased capacity announcement — source checked October 2, 2026
Fit detail
When Cerebras Inference is the right shortlist — and when it is not
Use these lists to kill bad comparisons early, before you compare logos.
Ideal for
- Shared API requests on the currently supported catalog
- Enterprise custom-weight serving on supported architectures
- Reserved-capacity workloads with measured latency and throughput requirements
Usually not ideal for
- Arbitrary unsupported architectures through Shared Inference
- Expecting shared API rates or model IDs to apply to dedicated custom weights
- Raw NVIDIA GPU rental through the Cerebras inference API
Spend & ops
How cost behaves — and what breaks first
No invented $/hour quotes. These notes explain the failure modes that show up on the first real invoice.
Cost mental model
Shared per-model tokens and enterprise reserved capacity have separate terms. No current numeric model tariff is verified here; confirm the exact rate and quota before buying credits. Hardware speed alone does not establish completed-task cost.
Ops notes
- Use a shared model ID or a dedicated endpoint ID as appropriate; an endpoint can be repointed, so pin an immutable version when required.
- Check per-tier context/rate limits and supported dedicated architectures before uploading custom weights.
- Activating a trial with a valid payment method gives a one-time $5 credit that expires after 30 days, without automatic paid enrollment. Access pauses at expiry/exhaustion unless you separately buy PayGo credits. Confirm the current model tariff before forecasting spend.
Fit scores
Cerebras Inference on the decision axes
Same editorial scale as the hub chart — useful for shortlists, not as a price quote.
Cerebras Inference
- GPU choice1/10
- Time-to-serving (editorial)5/10
- Price clarity4/10
- Production ops3/10
- Open-model breadth2/10
Among all 16 providers
GPU choice
Pick exact GPUs
Time-to-serving (editorial)
Warm-path fit
Price clarity
Easy to forecast
Production ops
Less DIY ops
Open-model breadth
Catalog depth
Scores are editorial planning ratings (1–10) for product shape — not published $/hour quotes or vendor SLAs. Verify current pricing on each provider site.
Related
Keep the decision attached to the rest of the stack
Cloud rental is one path. Local VRAM and open-weight fit still matter.