// head_to_head
Nemotron 3 Ultra vs Hy4 preview
Published LiveBench scores across all seven categories, live list pricing, context windows, and the measured cost of a point of capability — for both models, side by side.
Nemotron 3 Ultra is the cheaper of the two; neither can be ranked on quality here.
Hy4 preview does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.
nvidia
Nemotron 3 Ultra
- Blended / 1M
- $1.05
- Context
- 262K
- Released
- Jun 4, 2026
- Overall score
- 67.4
tencent
Hy4 preview
- Blended / 1M
- $1.25
- Context
- 1.0M
- Released
- Aug 28, 2026
- Overall score
- Not evaluated
Specs and pricing
| Metric | Nemotron 3 Ultra | Hy4 preview |
|---|---|---|
| LiveBench overall Mean of the seven LiveBench category scores, 0–100. Higher is better. | 67.4 | — |
| Cost per point Measured benchmark spend divided by overall score — dollars per point of capability. | $0.2118 | — |
| Blended price / 1M 3:1 input:output mix, the usual shape of production traffic. | $1.05win | $1.25 |
| Input price / 1M | $0.600win | $0.834 |
| Output price / 1M | $2.40 | $2.50 |
| Cached input / 1M Price of an input token served from the prompt cache, where the provider publishes one. | $0.120 | $0.042win |
| Context window | 262K | 1.0Mwin |
| Max output tokens | 183Kwin | 64K |
Benchmarks by category
An overall score averages away the thing you probably care about. These are the seven LiveBench categories scored separately — Nemotron 3 Ultra on top, Hy4 preview below, both out of 100.
What each one costs to run
Per-token prices are hard to feel. These are monthly list costs for both models across five workload shapes, using each provider's published cached-input rate where there is one.
| Workload | Nemotron 3 Ultra | Hy4 preview |
|---|---|---|
| Support chatbot 1.2K in / 400 out × 200K requests | $301.44/mo | $343.22/mo |
| RAG assistant 8K in / 600 out × 100K requests | $432.00/mo | $500.46/mo |
| Coding agent 40K in / 4K out × 20K requests | $403.20/mo | $423.76/mo |
| Document extraction 20K in / 1.5K out × 50K requests | $756.00/mo | $981.98/mo |
| Bulk classification 500 in / 20 out × 5M requests | $1,500/mo | $1,939/mo |
Which should you pick?
You need to fit large documents in one call
Hy4 preview
Wider context window — 1.0M against 262K.
You are cost-constrained
Nemotron 3 Ultra
Cheaper on blended list price at $1.05 per million tokens.
Nemotron 3 Ultra vs Hy4 preview FAQ
Which is better, Nemotron 3 Ultra or Hy4 preview?
Nemotron 3 Ultra is the cheaper of the two; neither can be ranked on quality here. Hy4 preview does not have a published LiveBench run, so this comparison covers price, context and declared capabilities only. A missing score means "not evaluated", not "worse" — the right way to separate these two is an eval on your own workload.
Is Nemotron 3 Ultra cheaper than Hy4 preview?
Nemotron 3 Ultra is cheaper. On a 3:1 input:output blend, Nemotron 3 Ultra lists at $1.05 per million tokens and Hy4 preview at $1.25 — Nemotron 3 Ultra is 19% cheaper. Input and output are priced separately — Nemotron 3 Ultra charges $0.600 in and $2.40 out, Hy4 preview charges $0.834 and $2.50 — so the model that looks cheaper flips depending on how output-heavy your workload is.
Does Nemotron 3 Ultra or Hy4 preview have a bigger context window?
Hy4 preview has the larger context window: 262K for Nemotron 3 Ultra against 1.0M for Hy4 preview. Note that a window you can fill is not a window you should fill — retrieval quality usually degrades well before the limit, and you pay for every token you put in it.
Do Nemotron 3 Ultra and Hy4 preview support prompt caching?
Both publish a cached-input rate: $0.120 per million for Nemotron 3 Ultra and $0.042 for Hy4 preview, against full input rates of $0.600 and $0.834. On a workload with a long stable prefix — a system prompt, a tool schema, a retrieved corpus — that changes the economics more than the headline price does.
Related comparisons
How these numbers are produced
- Price — provider list price from OpenRouter, refreshed every 15 minutes. “Blended” is a 3:1 input:output mix.
- Scores — LiveBench release 2026-06-25, using their own category map. Each model shows its strongest published run. A blank means “not evaluated”, never “bad”.
- Cost per point — the measured dollars LiveBench spent on the run, divided by the score it earned.
- “Win” — awarded only past a threshold: one full point on a benchmark score, 10% on a price, 25% on a context window. Anything tighter reports as a tie, because effort settings alone move a LiveBench score by more than that.
Published benchmarks rank models on someone else's tasks. Before committing, see LLM & agent evaluation for building an eval on your own.