Compare AI models
| Specifications | ||
ParametersThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. | 120B | 2.4T |
Context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider. | 1M | 262k |
| API pricingUSD per 1M tokens · lower wins | ||
Cheapest inputLowest input rate across every provider on OpenRouter, the lab included. The cheapest endpoint may run a quantised build or a shorter context — see "Available from" on the model page. | $0.00Nvidia | — |
Cheapest outputLowest output rate across every provider on OpenRouter, the lab included. May come from a different provider than the cheapest input. | $0.00Nvidia | — |
| $29.50Darkbloom · nvfp4 | — | |
| Benchmarks | ||
Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better. | 60.67% | 76.03% |
GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better. | 89.52% | 92.55% |
IFBenchInstruction following — Tests whether the AI can follow detailed instructions and satisfy multiple constraints at once. Higher is better. | 70.86% | 79.93% |
SWE-Bench ProAgentic coding — Can the AI fix real bugs in real software? It's handed actual problems from open-source projects and has to write code that genuinely solves them. Higher is better. | 51.78% | — |
SWE-Bench VerifiedCoding — Real coding tasks pulled from open-source projects — the AI has to find and fix actual bugs. A human-checked version of the original SWE-Bench. Higher is better. | 71.1% | — |
SWE-Bench MultilingualMultilingual coding — Like SWE-Bench, but the coding problems span many programming languages, not just one. Tests how broadly the AI can code. Higher is better. | 67.72% | — |
LiveCodeBenchCompetitive coding — Coding problems published so recently the AI can't have seen them in training — a contamination-free test of raw programming skill. Higher is better. | 88.22% | — |
Terminal-Bench 4.0Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Version 4.0 recalibrated how much time, CPU and memory each task gets, removed eight tasks and fixed nineteen, so fewer runs fail for reasons that have nothing to do with the model. Scores are not comparable with earlier versions. Higher is better. | 0.3% | — |
τ²-bench · retailAgentic tasks — task success in the retail domain of the second-generation customer-service evaluation | 69.24% | — |
τ²-bench · airlineAgentic tasks — task success in the airline domain of the second-generation customer-service evaluation | 76.5% | — |
τ²-bench · telecomAgentic tasks — task success in the telecom domain of the second-generation customer-service evaluation | 73.19% | — |
BrowseCompWeb browsing — Can the AI browse the web and track down hard-to-find answers? Higher is better. | 78.91% | — |
Humanity's Last Exam · no toolsMultidisciplinary reasoning — Humanity's Last Exam — extremely hard expert questions across many subjects, written so you can't just look up the answer. “No tools” means the AI answers on its own. Higher is better. | 29.2% | — |
Humanity's Last Exam · text only, no toolsMultidisciplinary reasoning — accuracy on the 2,158 text-only questions without tools | 31.79% | — |
Humanity's Last Exam · text only, with toolsMultidisciplinary reasoning — The text-only questions from Humanity's Last Exam, answered with access to tools such as search or code execution. The score is the percentage answered correctly. | 46.01% | — |
OSWorld-VerifiedAgentic computer use — Can the AI actually operate a computer — clicking, typing, and using real apps — to finish tasks on its own? Higher is better. | 51.14% | — |
CharXiv ReasoningChart reasoning — Can the AI read and reason about complex charts and figures, not just text? Higher is better. | 80.82% | — |
BabyVisionVisual reasoning — Tests core visual reasoning — seeing and understanding images the way even young children can, which AIs often find surprisingly hard. Higher is better. | 29.6% | — |
| Overview | ||
| Company | NVIDIA | Qwen |
| Release date | Oct 15 2026 | Aug 12 2026 |
| Access | Open Weight | Open Weight |
| Model details | View model | View model |
Other comparisons
Nemotron 3.5 Super VLvsClaude Fable 5.2Qwen3.8-2.4T-A95BvsClaude Fable 5.2Nemotron 3.5 Super VLvsGPT-6.2 AstraQwen3.8-2.4T-A95BvsGPT-6.2 AstraNemotron 3.5 Super VLvsGemini 4 NeonQwen3.8-2.4T-A95BvsGemini 4 NeonNemotron 3.5 Super VLvsMuse EmberQwen3.8-2.4T-A95BvsMuse EmberNemotron 3.5 Super VLvsGrok 4.7Qwen3.8-2.4T-A95BvsGrok 4.7Nemotron 3.5 Super VLvsDeepSeek-V4.1-Flash-1017Qwen3.8-2.4T-A95BvsDeepSeek-V4.1-Flash-1017Frequently asked questions
Qwen3.8-2.4T-A95B leads Nemotron 3.5 Super VL on 3 of the 3 benchmarks they both report (Terminal-Bench 2.1, GPQA Diamond, IFBench). Qwen3.8-2.4T-A95B shipped 64 days before Nemotron 3.5 Super VL, so benchmark comparisons should account for the intervening progress.
Nemotron 3.5 Super VL has 120B parameters, while Qwen3.8-2.4T-A95B has 2.4T. Context windows are 1M (Nemotron 3.5 Super VL) vs 262k (Qwen3.8-2.4T-A95B).
On Terminal-Bench 2.1, Qwen3.8-2.4T-A95B leads at 76.03% vs Nemotron 3.5 Super VL at 60.67%. On GPQA Diamond, Qwen3.8-2.4T-A95B leads at 92.55% vs Nemotron 3.5 Super VL at 89.52%. On IFBench, Qwen3.8-2.4T-A95B leads at 79.93% vs Nemotron 3.5 Super VL at 70.86%.