DeepInfra raises $107M Series B to scale the inference cloud — read the announcement

DeepStart x NVIDIA Inception

NVIDIA Inception

10,000,000,000 free tokens*

Usable on the NVIDIA Nemotron family

To qualify your startup needs to meet the following

Have raised between 250K and 10M USD

Founded in the last 5 years

NVIDIA Inception membership

Available in most countries — some regions are excluded by export regulations

* at NVIDIA Nemotron 3 Ultra cached-input prices ($0.10/M). Your grant is $1,000 in credits, spendable across the Nemotron family at list prices — output and uncached input draw it down faster. Valid for 12 months from approval.

Apply now

Stack with DeepStart

Approved Inception startups also receive the standard DeepStart grant — ~50,000,000,000 free tokens at DeepSeek V4 Flash prices, usable on any model, on top of the Nemotron credits. One application covers both.

Learn about DeepStart →

Featured Nemotron models

Your credits work across the NVIDIA Nemotron family

text-generation

nvidia/

NVIDIA-Nemotron-3.5-Super-VL

NVIDIA Nemotron 3.5 Super VL is an open multimodal model for agents that see, reason, and act. Its 120B-parameter Mixture-of-Experts design (12B active) understands text, images, documents, GUIs, and video, and adds reasoning and tool use on top.

ZDR

fp4

256k

$0.02 cached, $0.15 in, $0.50 out / 1M

text-generation

nvidia/

NVIDIA-Nemotron-3.5-Lightning

NVIDIA Nemotron 3.5 Lightning is NVIDIA's fastest open model for always-on agents and high-volume specialized tasks. It delivers a substantial leap in agentic capability over its predecessor Nemotron 3 Nano, with up to 4x higher throughput on a 1M-token context.

ZDR

256k

$0.03 cached, $0.06 in, $0.16 out / 1M

text-generation

nvidia/

Nemotron-Content-Safety-3.5

Nemotron Content Safety 3.5 is a multimodal safety classifier developed by NVIDIA. A compact safety model that handles text, images, and custom policies. It outputs a safe/unsafe classification plus a reasoning trace, and can be used as an inference-time guardrail, as a judge for LLM safety testing and evaluation, or with the accompanying training dataset to post-train models for safer behavior.

NVIDIA-Nemotron-3-Ultra-550B-A55B

Nemotron 3 Ultra is built for, frontier reasoning, orchestration, coding agents, deep research, and complex enterprise workflows. It delivers up to 5x faster inference and up to 30% lower cost for agentic workloads while supporting up to 1M token context.

$0.10 cached, $0.50 in, $2.20 out / 1M

text-generation

nvidia/

NVIDIA-Nemotron-3-Super-120B-A12B

NVIDIA Nemotron 3 Super is a hybrid Mixture-of-Experts (MoE) model engineered for highest compute efficiency and accuracy in multi-agent applications and specialized agentic systems. It is optimized to run many collaborating agents per application on a single GPU, delivering high accuracy for reasoning, tool use, and instruction following.

$0.075 in, $0.40 out / 1M

speech-recognition

nvidia/

NVIDIA-Nemotron-3.5-Transcribe-30B-A4B

Nemotron 3.5 Transcribe is an open-weight 30B-A4B speech-augmented language model from NVIDIA for offline transcription in 40 languages. It handles recordings up to one hour and adds speech translation, contextual and speaker-attributed transcription, word timestamps and audio question answering, leading open-weight models on English, long-form and multilingual ASR.

Nemotron-3.5-Rerank-8B-BF16

Nemotron-3.5-Rerank-8B-BF16 is an 8B-parameter reranking model from NVIDIA that scores the relevance of documents to a query, for reordering the candidates retrieved by an embedding model. It supports long inputs up to 32k tokens and pairs naturally with the Nemotron-3.5-Embed models in multilingual RAG pipelines.

Nemotron-3.5-Embed-8B-BF16

Nemotron-3.5-Embed-8B-BF16 is a multimodal embedding model from NVIDIA that maps text, images, and page screenshots into a shared 4096-dimensional vector space for visual document retrieval and semantic search. Built on the Nemotron-3-Embed-8B text encoder with a vision encoder, it supports up to 32k tokens and Matryoshka dimensions, and delivers the highest retrieval quality in the Nemotron 3.5 family.

ZDR

fp16

32k

$0.035 / 1M tokens · $0.070 / 1M image tokens

embeddings

nvidia/

Nemotron-3.5-Embed-1B-BF16

Nemotron-3.5-Embed-1B-BF16 is a compact multimodal embedding model from NVIDIA that maps text, images, and page screenshots into a shared 2048-dimensional vector space for visual document retrieval and semantic search. It pairs the Nemotron-3-Embed-1B text encoder with a vision encoder, supports up to 32k tokens and Matryoshka dimensions, and suits multilingual RAG over PDFs, slides, and scanned documents.

ZDR

fp16

32k

$0.015 / 1M tokens · $0.035 / 1M image tokens

speech-recognition

nvidia/

Nemotron-3-Diarization

Nemotron 3 Diarization is an open-weight streaming Sortformer model, engineered for low-latency speaker diarization across up to 8 speakers, ordered by arrival time. It powers meeting transcription, call analytics, and voice-agent pipelines—resolving who spoke when with generic speaker labels and timestamps in a single streaming pass, offline or in real time.

Nemotron-3-Embed-8B is a multilingual text embedding model from NVIDIA, based on Ministral-3-8B, that maps text into 4096-dimensional dense vectors for retrieval and semantic similarity. Spanning 34 languages, it targets multilingual RAG and question-answering over large corpora and achieves state-of-the-art results on the RTEB leaderboard.

Nemotron-3-Embed-1B-NVFP4

Nemotron-3-Embed-1B-NVFP4 is the NVFP4-quantized version of Nemotron-3-Embed-1B-BF16 — a multilingual text embedding model from NVIDIA that maps text into 2048-dimensional dense vectors for retrieval and semantic similarity. Optimized for NVIDIA Blackwell GPUs (e.g. RTX 6000 PRO, GB200), it retains near-BF16 quality (RTEB 72.0 vs 72.4) at a fraction of the memory and compute.

Apply for the NVIDIA Inception offer

Tell us about your startup. We review applications within a few business days; approved startups receive 10,000,000,000 free Nemotron tokens ($1,000 in credits) plus the standard DeepStart grant.