Serverless
$0
to start · pay per token
Shared, autoscaled endpoints for every catalogue model. Rate limits rise automatically with spend.
- $5 free credit, no card
- 600 RPM → 60,000 RPM
- Batch API at 50% off
- Community support
A fictional brand, designed and built by Get Ranked On Top. Not a real company.
All systems nominal · 14 of 14 regions serving
112 ms median time-to-first-token. 1,184 tokens per second on a 70B model. 60+ open-weight models behind one API, on GPUs we schedule ourselves across 14 regions — serverless, dedicated or inside your VPC.
In production at
HALDEN/LABStidewaterParallax▲LegalMOSAIC HEALTHquorum.aiLattice & LoomNORTHFORMbrevet01Playground
Pick a prompt. The readouts show what our edge measures on every request: time to first token, then sustained throughput.
// output streams here
// readouts are measured from request start02Model catalogue
Every model ships with tool calling, JSON mode and streaming. New open-weight releases go live within 48 hours, benchmarked before they are listed.
| Model | Type | Context | Input $/M | Output $/M | p50 TTFT | Throughput | Status |
|---|---|---|---|---|---|---|---|
| vela-3-405b | chat | 128k | $3.20 | $3.20 | 214 ms | 412 tok/s | serving |
| vela-3-70b | chat | 128k | $0.54 | $0.72 | 112 ms | 1,184 tok/s | serving |
| vela-3-8b | chat | 128k | $0.06 | $0.08 | 71 ms | 2,940 tok/s | serving |
| orrin-coder-34b | code | 64k | $0.38 | $0.48 | 131 ms | 1,026 tok/s | serving |
| lumen-vl-90b | vision | 32k | $0.90 | $1.10 | 188 ms | 604 tok/s | serving |
| tern-reason-32b | reasoning | 128k | $0.42 | $1.68 | 146 ms | 880 tok/s | serving |
| sable-embed-2 | embed | 8k | $0.04 | — | 18 ms | — | serving |
| sable-rerank-2 | rerank | 8k | $0.05 | — | 24 ms | — | serving |
Showing 8 of 62. Latency: client-measured, same-metro, 512-token prompt, 30-day p50.
03Benchmarks
vela-3-70b, identical weights and sampling settings on every provider. We publish the harness; run it yourself before you believe us.
04Quickstart
The SDKs follow the chat completions schema you already ship. Most migrations are a base URL and a model string.
from halvorn import Halvorn client = Halvorn() # reads HALVORN_API_KEY stream = client.chat.completions.create( model="vela-3-70b", messages=[{"role": "user", "content": "Ship it."}], max_tokens=512, stream=True, ) for chunk in stream: print(chunk.delta, end="", flush=True)
05Regions
Anycast routing sends every request to the nearest healthy GPU pool. Pin a key to a region, or to EU-only, for data residency.
| Region | Metro | RTT |
|---|---|---|
| sfo1 | San Francisco | 4 ms |
| pdx1 | Portland | 11 ms |
| ord1 | Chicago | 9 ms |
| yyz1 | Toronto | 12 ms |
| iad1 | Ashburn | 7 ms |
| gru1 | São Paulo | 18 ms |
| lhr1 | London | 6 ms |
| fra1 | Frankfurt | 5 ms |
| arn1 | Stockholm | 9 ms |
| bom1 | Mumbai | 14 ms |
| sin1 | Singapore | 8 ms |
| icn1 | Seoul | 10 ms |
| hnd1 | Tokyo | 6 ms |
| syd1 | Sydney | 13 ms |
06Pricing
No seat fees, no minimums on serverless, no egress charges. Cached input is 75% off; the batch API is 50% off with a 24-hour window.
| Model | Input | Cached input | Output | Batch input |
|---|---|---|---|---|
| vela-3-8b | $0.06 | $0.015 | $0.08 | $0.03 |
| vela-3-70b | $0.54 | $0.135 | $0.72 | $0.27 |
| vela-3-405b | $3.20 | $0.80 | $3.20 | $1.60 |
| orrin-coder-34b | $0.38 | $0.095 | $0.48 | $0.19 |
| tern-reason-32b | $0.42 | $0.105 | $1.68 | $0.21 |
| lumen-vl-90b | $0.90 | $0.225 | $1.10 | $0.45 |
| sable-embed-2 | $0.04 | — | — | $0.02 |
$0
to start · pay per token
Shared, autoscaled endpoints for every catalogue model. Rate limits rise automatically with spend.
$2.85
per GPU-hour · billed per second
Reserved capacity with your own weights, LoRA adapters and a private endpoint in any region.
Custom
annual commit
Halvorn's runtime in your VPC or data centre, with committed-use pricing and a named engineer.
07Enterprise
The same runtime that serves 41.7B tokens a day, deployed where your data already lives. Control plane with us, data plane with you.
Our scheduler and kernels deploy into your VPC via a single Terraform module. Weights, prompts and logs never cross your boundary.
Prompts and completions are held in memory for the life of the request, then dropped. No training on your traffic, ever — it is in the contract.
Customer-managed encryption keys, SSO with SCIM, per-key spend caps and an immutable audit log exported to your SIEM.
Capacity planning, kernel tuning for your sequence lengths, and a 15-minute P1 response window around the clock.
08Customers
From 3-person startups to teams pushing 2B+ tokens a day. Numbers are theirs.
“We moved 2.3 billion tokens a day across in a weekend. p95 latency on our support agent fell from 2.1 s to 640 ms, and the invoice fell by 41%.”
“Zero retention was the gate for our legal team. Halvorn put it in the MSA on the first call instead of the fourth.”
“Speculative decoding on our own fine-tune, in our VPC, without us writing a CUDA kernel. That was the whole pitch, and it held.”
09Talk to an engineer
Tell us your model, peak RPM and sequence lengths. We reply with a latency and cost estimate from a real benchmark run — usually inside a day.
10FAQ
Halvorn implements the de-facto chat completions, embeddings and batch schemas. Most teams change a base URL and a model string; streaming, tool calls and JSON mode behave identically.
p50 time-to-first-token is measured client-side from a request leaving a VM in the same metro to the first streamed byte, 512-token prompts, 30-day rolling window. We publish the raw percentiles daily at status.halvorn.example.
No. Serverless traffic is retained for 0 seconds by default; opt-in 30-day logging exists only for your own debugging and is encrypted with your key.
Yes. Upload safetensors or point us at a bucket. Dedicated endpoints serve full fine-tunes; serverless endpoints hot-swap up to 1,000 LoRA adapters per base model at no extra cost.
Requests fail over to the nearest healthy region in under 3 seconds unless you pin residency. EU-pinned keys only ever fail over between lhr1, fra1 and arn1.
Per million tokens on serverless, per GPU-second on dedicated, both invoiced monthly. Cached input tokens are billed at 25% of the input rate; the batch API is 50% off with a 24-hour window.