Beam 1.0 vs Nemotron 3.5 Super VL (120B A12B): Specs & Benchmark Comparison

Beam 1.0 is developed by Reflection AI, while Nemotron 3.5 Super VL (120B A12B) comes from NVIDIA. Both were released in October 2026. Beam 1.0 is the larger model at roughly 501 billion parameters, against 120 billion for Nemotron 3.5 Super VL (120B A12B).

The two models share 11 published benchmarks. Beam 1.0 leads on 10 of them, Nemotron 3.5 Super VL (120B A12B) on 1. The widest gaps are on Tau3 Banking, where Beam 1.0 scores 36.1% against 18.9%; SWE-Bench Pro, where Beam 1.0 scores 66.1% against 51.8%. Averaged across everything we track, Beam 1.0 sits at 61.6% and Nemotron 3.5 Super VL (120B A12B) at 60.9%.

Nemotron 3.5 Super VL (120B A12B) is the cheaper API at $0.20 per million input tokens and $1 per million output tokens, roughly 2 times cheaper than Beam 1.0 at $0.40 and $1.20. Beam 1.0 takes the larger context window at 1M tokens, compared with 262K for Nemotron 3.5 Super VL (120B A12B). Beam 1.0 accepts text as input, while Nemotron 3.5 Super VL (120B A12B) accepts text, images, and video.

CharacteristicBeam 1.0Nemotron 3.5 Super VL (120B A12B)
CompanyReflection AINVIDIA
Release DateOctober 16, 2026October 15, 2026
Parameters501B120B
MultimodalNoYes
Context (input)1.0M262K
Context (output)131K262K
Input Price / 1M$0.40$0.20
Output Price / 1M$1.20$1.00
Average Score61.6%60.9%
Benchmarks
Tau3 Banking36.1%18.9%
SWE-Bench Pro66.1%51.8%
Terminal-Bench 2.180.1%66.7%
SWE-bench Multilingual78.3%67.7%
SWE-Bench Verified81.6%71.1%
AA-LCR78.2%68.6%
IFBench79.6%70.9%
Humanity's Last Exam36.0%31.8%
SciCode49.2%47.1%
BrowseComp76.8%78.0%
GPQA90.6%89.5%

Visual Benchmark Comparison

Beam 1.0
Nemotron 3.5 Super VL (120B A12B)
Tau3 Banking0.4 vs 0.2
0.4
0.2
SWE-Bench Pro0.7 vs 0.5
0.7
0.5
Terminal-Bench 2.10.8 vs 0.7
0.8
0.7
SWE-bench Multilingual0.8 vs 0.7
0.8
0.7
SWE-Bench Verified0.8 vs 0.7
0.8
0.7
AA-LCR0.8 vs 0.7
0.8
0.7
IFBench0.8 vs 0.7
0.8
0.7
Humanity's Last Exam0.4 vs 0.3
0.4
0.3
SciCode0.5 vs 0.5
0.5
0.5
BrowseComp0.8 vs 0.8
0.8
0.8
GPQA0.9 vs 0.9
0.9
0.9

Verdict

Both models show equal results — the choice depends on your specific use case.

Overall Performance

Both models show comparable average scores: Beam 1.0 — 0.6, Nemotron 3.5 Super VL (120B A12B) — 0.6.

Programming

On SWE-Bench, both models are nearly equal: Beam 1.0 — 0.8, Nemotron 3.5 Super VL (120B A12B) — 0.7.

API Cost

Nemotron 3.5 Super VL (120B A12B) is 1.3x cheaper: input $0.20/1M vs $0.40/1M tokens.

Context Window

Beam 1.0 supports a larger context: 1M vs 262K tokens.

Recency

Both models were released around the same time: 10/16/2026 and 10/15/2026.

More About These Models

Related Comparisons

Frequently Asked Questions

Which is better for coding — Beam 1.0 or Nemotron 3.5 Super VL (120B A12B)?
On the SWE-Bench benchmark, Beam 1.0 shows a better result: 81.6% vs 71.1%.
Which model is cheaper — Beam 1.0 or Nemotron 3.5 Super VL (120B A12B)?
Nemotron 3.5 Super VL (120B A12B) is cheaper for input: $0.20 per 1M tokens vs $0.40.
Which has a larger context window — Beam 1.0 or Nemotron 3.5 Super VL (120B A12B)?
Beam 1.0 supports a larger context: 1,048,576 tokens vs 262,144.

The Beam 1.0 and Nemotron 3.5 Super VL (120B A12B) comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the Beam 1.0 or Nemotron 3.5 Super VL (120B A12B) page. See also the complete list of AI model comparisons.