NVIDIA Nemotron Labs Dory 1.0 vs Qwen 2.5 Max
A direct, empirical evaluation of NVIDIA Nemotron Labs Dory 1.0 (NVIDIA) against Qwen 2.5 Max (Alibaba Cloud).
NVIDIA Nemotron Labs Dory 1.0
Developed by NVIDIA · 4 Billion Parameters (Looped Transformer) parameters
Ultra-low VRAM memory residency (4.4 GB FP8), 256k long-context needle retrieval, and recurrent layer depth execution.
Qwen 2.5 Max
Developed by Alibaba Cloud · Proprietary (~700B) parameters
Cross-lingual document extraction and structured structured output generation.
| Metric | NVIDIA Nemotron Labs Dory 1.0 | Qwen 2.5 Max |
|---|---|---|
| SWE-bench Verified | 66.4% | 51.4% |
| MMLU-Pro | 82.8% | 77.2% |
| MATH-500 | 85.6% | 89.4% |
| Context Window | 256,000 tokens | 128,000 tokens |
| Input Pricing (per 1M) | $0.15 | $1.60 |
| Output Pricing (per 1M) | $0.45 | $6.40 |
Architectural & Practical Deployment Analysis
Evaluating NVIDIA Nemotron Labs Dory 1.0 (NVIDIA) alongside Qwen 2.5 Max (Alibaba Cloud) highlights key technical trade-offs across neural parameter scale and compute efficiency. NVIDIA Nemotron Labs Dory 1.0 executes across 4 Billion Parameters (Looped Transformer) total parameters with 4.1B Active Recurrent Block (Equivalent to 16B Depth) active per token, whereas Qwen 2.5 Max operates on Proprietary (~700B) parameters with Dense/MoE Hybrid active per token.
On SWE-bench Verified software engineering tasks, NVIDIA Nemotron Labs Dory 1.0 achieves 66.4% autonomous resolution rate, compared to 51.4% for Qwen 2.5 Max. On complex reasoning benchmarks, NVIDIA Nemotron Labs Dory 1.0 posts 82.8% on MMLU-Pro and 85.6% on MATH-500, against 77.2% and 89.4% respectively for Qwen 2.5 Max.
From an operational economics perspective, processing 10,000,000 mixed tokens (7M input / 3M output) costs approximately $2.40 on NVIDIA Nemotron Labs Dory 1.0 versus $30.40 on Qwen 2.5 Max.
Recommendation: Deploy NVIDIA Nemotron Labs Dory 1.0 when your infrastructure prioritizes ultra-low vram memory residency (4.4 gb fp8), 256k long-context needle retrieval, and recurrent layer depth execution.. Deploy Qwen 2.5 Max when workload requirements favor cross-lingual document extraction and structured structured output generation..