Model Comparison
Devstral Medium
vs. GPT-5.5 (xhigh)
Comparing 2 AI models · 12 benchmarks · Mistral, OpenAI
Recommended Pick
Op GPT-5.5 (xhigh) 11 metric wins Strongest on: Throughput, Reasoning, Intelligence
Best Value
Devstral Medium
100.0 value score
22.6 reasoning / $0.80/1M
Lowest Price
Devstral Medium
$0.40/1M input price
Best Reasoning
GPT-5.5 (xhigh)
66.0 reasoning score
Blends available reasoning benchmarks
Best for Coding
GPT-5.5 (xhigh)
59.1 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
Devstral Medium has the strongest quality-to-price mix at 100.0 out of 100 value points.
Price gap
Devstral Medium is 12.5x cheaper on input tokens than GPT-5.5 (xhigh).
Speed gap
GPT-5.5 (xhigh) generates about 1.4x as many tokens per second as Devstral Medium.
Reasoning gap
GPT-5.5 (xhigh) leads Devstral Medium by 43.4 points on reasoning.
Coding gap
GPT-5.5 (xhigh) leads Devstral Medium by 43.2 points on coding.
Response Face-Off
Run one prompt through the selected models and compare response quality with live speed and cost context.
Devstral Medium
Mistral
TTFT
—
Time
—
tok/s
—
Tokens
—
Cost
—
GPT-5.5 (xhigh)
OpenAI
TTFT
—
Time
—
tok/s
—
Tokens
—
Cost
—
Which answer was more useful?
Full Comparison
| Metric | Mi Devstral Medium | Top Pick Op GPT-5.5 (xhigh) |
|---|---|---|
| Pricing per 1M tokens | ||
| Input Cost | $0.40/1M | $5.00/1M |
| Output Cost | $2.00/1M | $30.00/1M |
| Blended (3:1) | $0.80/1M | $11.25/1M |
| Specifications | ||
| Organization | Mistral | OpenAI |
| Release Date | Jul 10, 2025 | Apr 23, 2026 |
| Performance & Speed | ||
| Throughput | 40.1 tok/s | 55.1 tok/s |
| TTFT | 514ms | 42416ms |
| Latency | 514ms | 42416ms |
| Composite Indices | ||
| Value Score | 100.0 | 20.7 |
| Reasoning Score | 22.6 | 66.0 |
| Intelligence | 18.7 | 60.2 |
| Coding | 15.9 | 59.1 |
| Math | 4.7 | — |
| Standard Benchmarks | ||
| GPQA | 49.2% | 93.5% |
| MMLU Pro | 70.8% | — |
| HLE | 3.8% | 44.3% |
| LiveCodeBench | 33.7% | — |
| MATH 500 | 70.7% | — |
| AIME 2025 | 4.7% | — |
| AIME (Original) | 6.7% | — |
| SciCode | 29.4% | 56.1% |
| LCR | 28.7% | 74.3% |
| IFBench | 29.9% | 75.9% |
| TAU-bench v2 | 19.9% | 93.9% |
| TerminalBench Hard | 9.1% | 60.6% |
Key Takeaways
Devstral Medium offers the best value at $0.40/1M, making it ideal for high-volume applications and cost-conscious projects.
GPT-5.5 (xhigh) has the strongest reasoning profile with a 66.0 reasoning score, combining the available reasoning-heavy benchmarks.
GPT-5.5 (xhigh) reaches a 59.1 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
Devstral Medium
- Cost-sensitive applications
- High-volume processing
GPT-5.5 (xhigh)
- Complex reasoning tasks
- Research & analysis
- Code generation
- Software development
