Model Comparison
M2.5
vs. Nemotron 3 Super 120B A12B (Reasoning)
Comparing 2 AI models · 7 benchmarks · MiniMax, NVIDIA
Recommended Pick
Mi M2.5 8 metric wins Strongest on: Reasoning, Intelligence, GPQA
Best Value
Nemotron 3 Super 120B A12B (Reasoning)
100.0 value score
41.5 reasoning / $0.41/1M
Lowest Price
M2.5
$0.30/1M input price
Best Reasoning
M2.5
45.9 reasoning score
Blends available reasoning benchmarks
Best for Coding
Nemotron 3 Super 120B A12B (Reasoning)
37.7 coding index
Composite Indices
Higher is better; speed and price are normalized
Standard Benchmarks
Only benchmarks with data are shown
Differences That Matter
Best value
Nemotron 3 Super 120B A12B (Reasoning) has the strongest quality-to-price mix at 100.0 out of 100 value points.
Speed gap
Nemotron 3 Super 120B A12B (Reasoning) generates about 1.4x as many tokens per second as M2.5.
Reasoning gap
M2.5 leads Nemotron 3 Super 120B A12B (Reasoning) by 4.3 points on reasoning.
Top-pick rationale
M2.5 wins 8 measurable categories, including Reasoning, Intelligence, GPQA, SciCode.
Response Face-Off
Run one prompt through the selected models and compare response quality with live speed and cost context.
M2.5
MiniMax
TTFT
—
Time
—
tok/s
—
Tokens
—
Cost
—
Nemotron 3 Super 120B A12B (Reasoning)
NVIDIA
TTFT
—
Time
—
tok/s
—
Tokens
—
Cost
—
Which answer was more useful?
Full Comparison
| Metric | Top Pick Mi M2.5 | NV Nemotron 3 Super 120B A12B (Reasoning) |
|---|---|---|
| Pricing per 1M tokens | ||
| Input Cost | $0.30/1M | $0.30/1M |
| Output Cost | $1.20/1M | $0.75/1M |
| Blended (3:1) | $0.52/1M | $0.41/1M |
| Specifications | ||
| Organization | MiniMax | NVIDIA |
| Release Date | Feb 12, 2026 | Mar 11, 2026 |
| Performance & Speed | ||
| Throughput | 224.4 tok/s | 314.7 tok/s |
| TTFT | 6118ms | 1011ms |
| Latency | 15030ms | 7366ms |
| Composite Indices | ||
| Value Score | 86.8 | 100.0 |
| Reasoning Score | 45.9 | 41.5 |
| Intelligence | 33.7 | 25.4 |
| Coding | — | 37.7 |
| Standard Benchmarks | ||
| GPQA | 84.8% | 80.0% |
| HLE | 19.1% | 19.2% |
| SciCode | 42.6% | 36.0% |
| LCR | 66.0% | 60.0% |
| IFBench | 71.6% | 71.5% |
| TAU-bench v2 | 95.3% | 67.8% |
| TerminalBench Hard | 34.8% | 28.8% |
Key Takeaways
M2.5 offers the best value at $0.30/1M, making it ideal for high-volume applications and cost-conscious projects.
M2.5 has the strongest reasoning profile with a 45.9 reasoning score, combining the available reasoning-heavy benchmarks.
Nemotron 3 Super 120B A12B (Reasoning) reaches a 37.7 coding index, making it the top choice for software development and code generation tasks.
All models support context windows of ∞+ tokens, suitable for processing lengthy documents and maintaining extended conversations.
When to Choose Each Model
M2.5
- Cost-sensitive applications
- High-volume processing
- Complex reasoning tasks
- Research & analysis
Nemotron 3 Super 120B A12B (Reasoning)
- Code generation
- Software development
