GPT-6 Astra vs Nemotron 3 Nano Omni 30B-A3B

GPT-6 Astra leads on the Visual Intelligence Index by 21.9 points, and GPT-6 Astra scores higher on 7 of the 7 benchmarks where the two differ.

Model

Compared with

At a glance

Measure	GPT-6 Astra	Nemotron 3 Nano Omni 30B-A3B
Visual Intelligence Index	82.2%	60.3%
Rank	#1 of 9	#8 of 9
Cost per taskMean over the benchmarks, at list price.	$0.864	—
Latency per taskEach benchmark's median, averaged.	9.6s	14.7s
Tokens per task	86,253	35,341
Input / 1M tokens	$10.00	—
Output / 1M tokens	$50.00	—
Context window	1.05M	262K
Availability	Proprietary	Open weights

Each capability index is a weighted mean of the benchmarks that contribute to the capability.

Capability	GPT-6 Astra	Nemotron 3 Nano Omni 30B-A3B	Lead
Perception	80.8%	61.0%	+19.9
Spatial Reasoning	84.6%	62.0%	+22.5
Temporal Reasoning	83.0%	60.2%	+22.8
Action / Event Understanding	79.2%	55.6%	+23.7
Causal Reasoning	83.8%	63.6%	+20.2
Information Retrieval	80.7%	57.1%	+23.6

Benchmark	GPT-6 Astra	Nemotron 3 Nano Omni 30B-A3B	Lead
Video-MME v2	73.4%	35.0%	+38.5
LVBench	81.6%	54.2%	+27.4
Perception Test	92.5%	70.3%	+22.3
NExT-QA	90.7%	81.2%	+9.5
Q-Bench Video	68.3%	64.9%	+3.5
EgoSchema	83.4%	65.8%	+17.6
UCF101-AD	61.4%	31.0%	+30.4

3.5/ 5

Reachable1.0: There is an endpoint you can call without hosting anything.
First-party API — supported
Third-party API — supported
Portable0.0: You can run it yourself, and are not tied to one vendor.
Open weights — not supported
Self-hostable — not supported
Multimodal input0.5: It takes the footage directly, rather than frames you extracted.
Video ingestion — not supported
Audio — supported
Programmable1.0: Output you can parse, and tools it can call on its own.
Structured output — supported
Tool calling — supported
Operable1.0: Usable interactively and in bulk, not only one call at a time.
Streaming — supported
Batch API — supported

5.0/ 5

Reachable1.0: There is an endpoint you can call without hosting anything.
First-party API — supported
Third-party API — supported
Portable1.0: You can run it yourself, and are not tied to one vendor.
Open weights — supported
Self-hostable — supported
Multimodal input1.0: It takes the footage directly, rather than frames you extracted.
Video ingestion — supported
Audio — supported
Programmable1.0: Output you can parse, and tools it can call on its own.
Structured output — supported
Tool calling — supported
Operable1.0: Usable interactively and in bulk, not only one call at a time.
Streaming — supported
Batch API — supported