Skip to main content
⚡ EfficientBudget-Friendly

Devstral 2

by Mistral AI

Mistral's dedicated coding model with excellent code generation

Try Mistral AI

Input Price

$0.4

per 1M tokens

Output Price

$2

per 1M tokens

Context Window

262.144K

tokens (32.768K max output)

Specifications

Release Date2025-12-09
Category⚡ Efficient
Context Window262,144 tokens
Max Output32,768 tokens
ProviderMistral AI
Input Cost$0.4 / 1M tokens
Output Cost$2 / 1M tokens
Price TierBudget-Friendly

Capabilities

textcode

Related Use Cases

Monthly Cost Estimates

Usage LevelDaily TokensDailyMonthlyYearly
Light~50 requests/day100K in / 20K out$0.08$2.40$29
Medium~200 requests/day500K in / 100K out$0.40$12.00$146
Heavy~1K requests/day2,000K in / 500K out$1.80$54.00$657
Enterprise~5K requests/day10,000K in / 2,000K out$8.00$240.00$2920

Cost Calculator

Alternatives to Devstral 2

Frequently Asked Questions

How much does Devstral 2 API cost per million tokens?
Devstral 2 costs $0.4 per million input tokens and $2 per million output tokens as of 2026. These are the standard API rates from Mistral AI.
What is the Devstral 2 context window?
Devstral 2 supports a 262.144K context window (262,144 tokens), which means you can process up to 262.144K tokens in a single API call.
How much does Devstral 2 cost per month?
At medium usage (~200 requests/day with 500K input and 100K output tokens/day), Devstral 2 costs approximately $12.00/month. Light usage runs about $2.40/month, and heavy usage (~1K requests/day) around $54.00/month.
Is there a cheaper alternative to Devstral 2?
Yes — GPT-5 mini by OpenAI is a cheaper option at $0.25/M input tokens vs $0.4/M for Devstral 2. Other budget alternatives include models in the efficient tier.
Is Devstral 2 good for ai chatbot?
Devstral 2 is a efficient model with support for text, code. For ai chatbot, it offers 262.144K context and costs $0.4/M input tokens — a budget-friendly choice for this use case.

Devstral 2 Comparisons

Ready to use Devstral 2?

Get started with Mistral AI's API — free tier available for most models.

Try Mistral AI API →