Hy4 preview.
Hy4 preview is Tencent's Apache 2.0 mixture-of-experts model, 770B total and 49B active, with a 1M-token context; an early release aimed at engineering and analysis work.
8× Nvidia A100 SXM4 · $5.73/hr.
Most-aggressive quantisation we have a working recommendation for. Lower precision = less VRAM = cheaper hardware, at a small accuracy cost.
Cheapest hosted endpoints.
Cheapest hardware per quantisation.
Each row is one quantisation tier (the same weights compressed differently). Lower precision → lower VRAM → cheaper hardware, at the cost of small accuracy loss. $/hr is the cheapest rate on offer now. A multi-card setup uses a real instance of that size when one is rented; otherwise it is the single-card rate times the count.
| Quantisation | Cheapest GPU config | Total VRAM | Live $/hr | tokens/sec | |
|---|---|---|---|---|---|
|
FP16
FP16 — half precision (default)
|
2048 GB |
$24.57/hr
Cyfuture AI · on-demand
|
— | Compare → | |
|
FP8
FP8 — 8-bit float (Hopper / Blackwell)
|
1024 GB |
$12.41/hr
Cyfuture AI · on-demand
|
— | Compare → | |
|
INT4
INT4 — 4-bit integer (~4× VRAM saving)
|
640 GB |
$5.73/hr
est. 8 × single-card rate
Nosana · on-demand
|
— | Compare → |
What it costs per month across providers.
Estimate your monthly bill for Hy4 preview across every host that publishes per-token pricing. Slide your token volumes; the chart + table re-rank cheapest-first.
Cheapest provider on the left.
Total monthly cost — input + output tokens combined.
Bill breakdown.
Rent the GPU instead of paying per token.
For an open-weights model like Hy4 preview, you can rent a GPU and serve inference yourself. The math: cheapest GPU rental × 730 hours/month. Power is included in a rental's hourly price.
Assumes the GPU runs 24/7 at ~85% utilisation. If your traffic is bursty, you'll pay less for the API and probably more for the GPU (idle hours still cost rental). The breakeven analysis lives on the Self-host vs API breakeven tool.
About Hy4 preview.
Hy4 preview is an open-weights mixture-of-experts model from Tencent's Hy team, with 770B total parameters and 49B active per token, under Apache 2.0. Its context length is 1M tokens, and an extra layer supports speculative decoding. Tencent built its training data around software engineering, office and analysis work, game development and scientific research. It calls this an early version, with known issues such as reasoning longer than needed and over-verifying its own work.