Advertise here
#81

Nemotron 3.5 Super VL

NVIDIA Release: 2026-10-15 Tested on: 2026-10-16 00:19 nvidia/nemotron-3.5-super-vl::high
120B total (12B active)MoEOpen source
(high) (medium) (none)

Summary

Nemotron 3.5 Super VL scores 8.4 on AI BENCHY and ranks #81. It has 9.9 reliability, a 82.6% pass rate, $0.349 total cost, and 49.88s average response time.

What makes Nemotron 3.5 Super VL unique: It uses unusually many reasoning tokens, which can help explain its slower or more expensive runs.

Model facts

Researched on 2026-10-15

Reported
Parameters
120B total (12B active)
Architecture
MoE
Availability
Open source
License
OpenMDW-1.1

Hybrid Mamba-2/Transformer MoE with a C-RADIO vision encoder. NVIDIA publishes weights, implementation, training data and recipes under the permissive OpenMDW-1.1 license. The model card documents reasoning off, medium and full; this route is the paid exact ID, not the free alias.

Parameter source Availability source

Score

8.4

Consistency

8.0

Reliability

9.9

Total Cost (Current Price)

$0.349

Total Output Tokens

608,283

Total Input Tokens

296,001

Input Price

$0.150 / 1M

Output Price

$0.500 / 1M

Cache Read Price

$0.020 / 1M

Cache Write Price

N/A

Cache prices apply to input tokens. Reads reuse cached prompts; writes store them and can cost extra. Output tokens use the output price.

Tests Correct

Wrong Tests: 8

Attempt pass rate: 82.6%

Flaky tests

6

Flaky tests had mixed outcomes across runs (at least one pass and one fail).

Response Time (avg)

49.88s

Response Time (max): 310.51s

Response Time (total): 1147.24s

Wrong answer: 3 Did not follow instructions: 2 No answer: 2 API error: 1

Hamster playing table tennis

Prompt: Create a detailed SVG illustration of a hamster playing table tennis.

#81 Nemotron 3.5 Super VL

high
Cost
$0.005
Time
25.6s
Tokens
9,076 tok

Charts

Choose the first model, then click a second model to open a side-by-side page.

Total Output Tokens

Score vs Total Output Tokens

Quick Compare

Nemotron 3.5 Super VLhighFree AvailablevsClaude Haiku 5.5maxNemotron 3.5 Super VLhighFree AvailablevsQwen3.8 2.4T A95BlowNemotron 3.5 Super VLhighFree AvailablevsQwen3.7 MaxmediumNemotron 3.5 Super VLhighFree AvailablevsGPT-5 MinimediumNemotron 3.5 Super VLhighFree AvailablevsClaude Sonnet 5mediumNemotron 3.5 Super VLhighFree AvailablevsMuse Glimmer 30BxhighNemotron 3.5 Super VLhighFree AvailablevsMuse Spark 1.3lowNemotron 3.5 Super VLhighFree AvailablevsGemini 3.1 Pro PreviewmediumNemotron 3.5 Super VLhighFree AvailablevsGrok 4.6highNemotron 3.5 Super VLhighFree AvailablevsInklinghighFree Available

Category Breakdown

Category Score Consistency Tests Correct
Agentic 10.0 10.0
Anti-AI Tricks 10.0 10.0
Coding 8.4 7.4
Combined 7.3 5.8
Data parsing and extraction 10.0 10.0
Domain specific 4.1 4.4
General Intelligence 6.5 3.4
Instructions following 8.5 6.8
Puzzle Solving 10.0 10.0
Tool Calling 10.0 10.0
Trivia 3.0 10.0

Compared models