NVIDIA Nemotron Labs Dory 1.0 vs Qwen 4 Max (Leaked Staging)
A direct, empirical evaluation of NVIDIA Nemotron Labs Dory 1.0 (NVIDIA) against Qwen 4 Max (Leaked Staging) (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 4 Max (Leaked Staging)
Developed by Alibaba Cloud · 1.2 Trillion (Sparse MoE) parameters
Repository-scale software engineering outscoring GPT-6 Sol, multilingual enterprise agents across 120 languages, and low-cost 1M-token processing.
| Metric | NVIDIA Nemotron Labs Dory 1.0 | Qwen 4 Max (Leaked Staging) |
|---|---|---|
| SWE-bench Verified | 66.4% | 78.6% |
| MMLU-Pro | 82.8% | 91.4% |
| MATH-500 | 85.6% | 95.6% |
| Context Window | 256,000 tokens | 1,000,000 tokens |
| Input Pricing (per 1M) | $0.15 | $1.20 |
| Output Pricing (per 1M) | $0.45 | $4.80 |
Architectural & Practical Deployment Analysis
Evaluating NVIDIA Nemotron Labs Dory 1.0 (NVIDIA) alongside Qwen 4 Max (Leaked Staging) (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 4 Max (Leaked Staging) operates on 1.2 Trillion (Sparse MoE) parameters with 96 Billion Active Parameters active per token.
On SWE-bench Verified software engineering tasks, NVIDIA Nemotron Labs Dory 1.0 achieves 66.4% autonomous resolution rate, compared to 78.6% for Qwen 4 Max (Leaked Staging). On complex reasoning benchmarks, NVIDIA Nemotron Labs Dory 1.0 posts 82.8% on MMLU-Pro and 85.6% on MATH-500, against 91.4% and 95.6% respectively for Qwen 4 Max (Leaked Staging).
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 $22.80 on Qwen 4 Max (Leaked Staging).
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 4 Max (Leaked Staging) when workload requirements favor repository-scale software engineering outscoring gpt-6 sol, multilingual enterprise agents across 120 languages, and low-cost 1m-token processing..