Nemotron Cascade 2 30B A3B
Mar 2026
Nemotron Cascade 2 30B A3BCompare inon
Benchmarks of Nemotron Cascade 2 30B A3B13
BenchmarkByScoreRank
Overall1
Knowledge4
Science2
Coding1
Agents1
Documents1
Legacy · saturated or no longer reported3
No benchmark matches
—per task?
self-host onlyprice
—output speed?
1Mcontext
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunalow+84%—$0.005Function calling only with none · Chat Completions+3
OpenQwen3.8 27Bxhigh×2.9—$1.01Context: 1M → 262K+5BetterGPT-5.6 Lunalow+80%—$0.01
BetterMiMo v2.5thinkingOpen+87%—$0.019Memory (4-bit): 19 GB → 186 GB
BetterClaude Haiku 5.5low×2.5—$0.024Remove sampling parameters+5ModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunalow+84%—$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow+80%—$0.01MiMo v2.5thinkingOpen+87%—$0.019Memory (4-bit): 19 GB → 186 GBClaude Haiku 5.5low×2.5—$0.024Remove sampling parameters+5GPT-5 Minihigh+44%—$0.053Context: 1M → 400KModelScoreCostPer taskWhat changes
Qwen3.8 27BxhighOpen×2.9—$1.01Context: 1M → 262K+5IK2 Horizon MoVA 36B A4BthinkingOpen×2.2—$1 / $4Context: 1M → 524K+1Qwen 3.5 27BthinkingOpen+96%—$0.30 / $2.40Context: 1M → 262KAG9v3 39A5BOpen+87%——Context: 1M → 131K+1Qwen3.6 27BthinkingOpen+84%—$0.62Context: 1M → 262KModelScoreCostPer taskWhat changes
Nemotron 3.5 Super VLOpen×2.3—$0.19Memory (4-bit): 19 GB → 75 GBNemotron 3 UltrathinkingOpen+96%—$0.58Memory (4-bit): 19 GB → 336 GBNemotron 3.5 Bolt 74B-A9BOpen+83%——Context: 1M → 256K+1Nemotron 3.5 LightningthinkingOpen+10%—$0.079Memory (4-bit): 19 GB → 20 GBNemotron 3 Super 120B A12BOpen+10%—$1.64Memory (4-bit): 19 GB → 74 GBNemotron 3 Nano Omni 30B A3B ReasoningOpen-12%—$0 / $0Context: 1M → 256K+1Llama Nemotron Super 49B v1.5non-thinkingOpen-37%—$0.40 / $0.40Context: 1M → 128K+1Llama 3.1 Nemotron Nano 4B v1.1Open-37%——Context: 1M → 128KAll NVIDIA models →Run it yourself
estimateFits on✗ 8 GB (phone, laptop)✗ 16 GB✓ 24 GB GPU · compressed✓ 48 GB GPU · 8-bit✓ 80 GB GPU · full quality
Licencenvidia-open-model-licensecommercial use with conditions
Benchmarks
all categoriesChat & Knowledge
Humanity's Last Exam12#162
AA Omniscience Index-51.5#217
CritPt (physics)0.6#141
AA Intelligence Index 4.3.2≈11.7#187
AA Omniscience · Accuracy18.3#180
AA-LCR (long context) 1.139.3#201
GPQA Diamond75.8#148
Coding
Terminal-Bench 2.120.6#112
Agents & Tool Use
τ²-bench · AA53.2#108
Writing & Language
IFBench (instructions)80.4#5
FAQ
How much does Nemotron Cascade 2 30B A3B cost?›
There is no per-unit API price: self-host only.
Is Nemotron Cascade 2 30B A3B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Nemotron Cascade 2 30B A3B, and what is it best at?›
#161 of 334 in Chat & Knowledge, #88 of 239 in Coding, #122 of 175 in Agents & Tool Use, #5 of 245 in Writing & Language. Strongest at: Instructions.
Is Nemotron Cascade 2 30B A3B better than MiniMax M1 80k?›
On AA Intelligence Index 4.3.2, Nemotron Cascade 2 30B A3B scores 11.7 vs 11.7 — about the same.
What is the context window of Nemotron Cascade 2 30B A3B?›
1M tokens.
When was Nemotron Cascade 2 30B A3B released? Is there a newer version?›
Released 2026-03-19. It is the current version in its line.
Is Nemotron Cascade 2 30B A3B open source?›
Yes, the weights are open (nvidia-open-model-license licence).
Can I run Nemotron Cascade 2 30B A3B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Nemotron Cascade 2 30B A3B read images?›
No, it accepts text only.
What is Nemotron Cascade 2 30B A3B?›
Nemotron Cascade 2 30B A3B is an open-weights language model from NVIDIA, released Mar 2026.
What are the best alternatives to Nemotron Cascade 2 30B A3B?›
GPT-6 Luna (better), Qwen3.8 27B (open weights).