NVIDIA: Nemotron 3 Ultra
budgetnvidia/nemotron-3-ultra-550b-a55b
NVIDIA Nemotron 3 Ultra is an open frontier-reasoning and orchestration model from NVIDIA, with 55B active parameters out of 550B total (MoE). Built on a hybrid Transformer-Mamba mixture-of-experts architecture, it...
AA INDEX22.9
CONTEXT262K
MODALITYtext->text
OPENNESSProprietary
ADDEDAug 2026
LAST PRICED18 min ago
Input
$0.60 /M
Output
$2.40 /M
Blended (40/60)
$1.68 /M
Cost / IQ
$2.9309 /M
Cost / IQ = Blended Price ÷ (AA Intelligence Score ÷ 40). Normalises price by model intelligence so you can compare value across tiers. Lower = better value per unit of intelligence. Not a transactional price.
24h0%
7d0%
Price History
30-day view, daily close
Low: $1.68/M
High: $1.68/M
Average: $1.68/M
Pricing Variants
Cached Input
$0.12/M
80% cheaper than standard
Batch (async)
$2.40/M
-42.9% cheaper than standard
In: $0.60/M · Out: $3.60/M
Provider Comparison
4 providers · click a column to sort
| Provider | Input $/M | Output $/M | Blended $/M ↑ | vs Median |
|---|---|---|---|---|
| ↓DeepInfra | $0.50 | $2.20 | $1.52 | -9.5% |
| BaseTen | $0.60 | $2.40 | $1.68 | median |
| Fireworks | $0.60 | $2.40 | $1.68 | median |
| Venice | $0.63 | $3.13 | $2.13 | +26.5% |
Blended price = 40% input + 60% output. Median price shown in header is the reference price for this model.
Cost / IQ Comparison
Cost / IQ (quality-adjusted $/M)$2.9309/M
Cost per GPT-4-equivalent token. Lower = better value. Comparable across all models.
48% above SIT TPI
Value Ranking (by Cost / IQ)
Rank #30 of 189 budget models
← YOU ARE HERE | NVIDIA: Nemotron 3 Ultra
Data Sources
Source
Aggregator & direct provider data
4 sources · updated recently
API Access
# Get this model's current price
curl https://api.inferenceindexer.ai/v1/models/nvidia/nemotron-3-ultra-550b-a55b
# Response (truncated)
{ "model_id": "nvidia/nemotron-3-ultra-550b-a55b", "name": "NVIDIA: Nemotron 3 Ultra", "blended_price_per_m": 1.68, "sit_adjusted_price": 2.930925 }