Model Comparison
V4 Pro (Reasoning, High Effort)
vs. Medium 3
Comparing 2 AI models · 12 benchmarks · DeepSeek, Mistral
Recommended Pick
De V4 Pro (Reasoning, High Effort) 14 metric wins Strongest on: Value, Output price, Blended price
Best Value
V4 Pro (Reasoning, High Effort)
100.0 value score
57.9 reasoning / $0.54/1M
Lowest Price
Medium 3
$0.40/1M input price
Best Reasoning
V4 Pro (Reasoning, High Effort)
57.9 reasoning score
Blends available reasoning benchmarks
Best for Coding
V4 Pro (Reasoning, High Effort)
43.2 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
V4 Pro (Reasoning, High Effort) has the strongest quality-to-price mix at 100.0 out of 100 value points.
Price gap
Medium 3 is 1.1x cheaper on input tokens than V4 Pro (Reasoning, High Effort).
Speed gap
V4 Pro (Reasoning, High Effort) generates about 1.6x as many tokens per second as Medium 3.
Reasoning gap
V4 Pro (Reasoning, High Effort) leads Medium 3 by 18.5 points on reasoning.
Coding gap
V4 Pro (Reasoning, High Effort) leads Medium 3 by 29.6 points on coding.
Response Face-Off
Run one prompt through the selected models and compare response quality with live speed and cost context.
V4 Pro (Reasoning, High Effort)
DeepSeek
TTFT
—
Time
—
tok/s
—
Tokens
—
Cost
—
Medium 3
Mistral
TTFT
—
Time
—
tok/s
—
Tokens
—
Cost
—
Which answer was more useful?
Full Comparison
| Metric | Top Pick De V4 Pro (Reasoning, High Effort) | Mi Medium 3 |
|---|---|---|
| Pricing per 1M tokens | ||
| Input Cost | $0.43/1M | $0.40/1M |
| Output Cost | $0.87/1M | $2.00/1M |
| Blended (3:1) | $0.54/1M | $0.80/1M |
| Specifications | ||
| Organization | DeepSeek | Mistral |
| Release Date | Apr 24, 2026 | May 7, 2025 |
| Performance & Speed | ||
| Throughput | 74.2 tok/s | 47.1 tok/s |
| TTFT | 1061ms | 531ms |
| Latency | 27902ms | 531ms |
| Composite Indices | ||
| Value Score | 100.0 | 46.3 |
| Reasoning Score | 57.9 | 39.5 |
| Intelligence | 49.8 | 18.8 |
| Coding | 43.2 | 13.6 |
| Math | — | 30.3 |
| Standard Benchmarks | ||
| GPQA | 90.5% | 57.8% |
| MMLU Pro | — | 76.0% |
| HLE | 33.5% | 4.3% |
| LiveCodeBench | — | 40.0% |
| MATH 500 | — | 90.7% |
| AIME 2025 | — | 30.3% |
| AIME (Original) | — | 44.0% |
| SciCode | 46.4% | 33.1% |
| LCR | 65.0% | 28.0% |
| IFBench | 71.3% | 39.3% |
| TAU-bench v2 | 94.2% | 24.3% |
| TerminalBench Hard | 41.7% | 3.8% |
Key Takeaways
Medium 3 offers the best value at $0.40/1M, making it ideal for high-volume applications and cost-conscious projects.
V4 Pro (Reasoning, High Effort) has the strongest reasoning profile with a 57.9 reasoning score, combining the available reasoning-heavy benchmarks.
V4 Pro (Reasoning, High Effort) reaches a 43.2 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
V4 Pro (Reasoning, High Effort)
- Complex reasoning tasks
- Research & analysis
- Code generation
- Software development
Medium 3
- Cost-sensitive applications
- High-volume processing
