Skip to content

Qwen3.5-35B-A3B vs Nemotron 3 Nano Omni 30B-A3B

Qwen3.5-35B-A3B leads on the Visual Intelligence Index by 5.4 points, and Qwen3.5-35B-A3B scores higher on 6 of the 7 benchmarks where the two differ.

At a glance

MeasureQwen3.5-35B-A3BNemotron 3 Nano Omni 30B-A3B
Visual Intelligence Index65.7%60.3%
Rank#7 of 9#8 of 9
Cost per taskMean over the benchmarks, at list price.——
Latency per taskEach benchmark's median, averaged.14.3s14.7s
Tokens per task25,75535,341
Input / 1M tokens——
Output / 1M tokens——
Context window256K262K
AvailabilityOpen weightsOpen weights

How cost and latency are measured

Capabilities

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

CapabilityQwen3.5-35B-A3BNemotron 3 Nano Omni 30B-A3B
Perception65.3%61.0%
Spatial Reasoning65.8%62.0%
Temporal Reasoning65.5%60.2%
Action / Event Understanding62.2%55.6%
Causal Reasoning70.1%63.6%
Information Retrieval63.6%57.1%

Benchmarks

BenchmarkQwen3.5-35B-A3BNemotron 3 Nano Omni 30B-A3B
Video-MME v242.5%35.0%
LVBench64.4%54.2%
Perception Test66.7%70.3%
NExT-QA83.4%81.2%
Q-Bench Video66.9%64.9%
EgoSchema77.2%65.8%
UCF101-AD41.3%31.0%

Usability

Qwen3.5-35B-A3B

4.5/ 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
Operable0.5

Usable interactively and in bulk, not only one call at a time.

  • Streaming — supported
  • Batch API — not supported

Nemotron 3 Nano Omni 30B-A3B

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