Skip to content

Commit 6a914fc

Browse files
committed
Support for NVIDIA Nemotron 3 Nano Omni
1 parent 1074b3b commit 6a914fc

2 files changed

Lines changed: 14 additions & 1 deletion

File tree

pyproject.toml

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
[project]
22
name = "kady"
3-
version = "0.3.6"
3+
version = "0.3.7"
44
description = "Kady — K-Dense BYOK Agent"
55
readme = "README.md"
66
requires-python = ">=3.13"

web/src/data/models.json

Lines changed: 13 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -352,6 +352,19 @@
352352
"modality": "text->text",
353353
"description": "Nemotron 3 Super is NVIDIA's 120B-parameter open hybrid Mixture-of-Experts model that activates only 12B parameters during inference, built on a hybrid Mamba-Transformer architecture with multi-token prediction. It features a 1M token context window and delivers over 50% higher token generation throughput compared to leading open models. Trained with multi-environment reinforcement learning across 10+ environments, it excels at complex multi-agent applications with long-term agent coherence and multi-step task planning. Fully open with weights, datasets, and recipes available under the NVIDIA Open License."
354354
},
355+
{
356+
"id": "openrouter/nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free",
357+
"label": "Nemotron 3 Nano Omni (free)",
358+
"provider": "NVIDIA",
359+
"tier": "budget",
360+
"context_length": 256000,
361+
"pricing": {
362+
"prompt": 0.0,
363+
"completion": 0.0
364+
},
365+
"modality": "text+image+video+audio->text",
366+
"description": "NVIDIA Nemotron\u2122 3 Nano Omni is a 30B-A3B open multimodal model designed to function as a perception and context sub-agent in enterprise agent systems. It accepts text, image, video, and audio inputs and produces text output, enabling agents to perceive and reason across modalities in a single inference loop.\n\nBuilt on a hybrid MoE Transformer-Mamba architecture with Conv3D video layers and Efficient Video Sampling (EVS), it delivers approximately 2\u00d7 higher throughput and 2.5\u00d7 lower compute for video reasoning versus separate vision + speech pipelines. It supports up to 300K context length and a 16,384 reasoning budget, with extended thinking enabled via reasoning.enabled on OpenRouter."
367+
},
355368
{
356369
"id": "openrouter/minimax/minimax-m2.1",
357370
"label": "MiniMax M2.1",

0 commit comments

Comments
 (0)