Devstral Medium

Devstral Medium is a text model from Mistral AI with a context window of 256K tokens and max output of 256K tokens. Pricing starts at 0.40 per million input tokens and 2.00 per million output tokens.

Capabilities

Vision Function Calling Reasoning JSON Schema System Messages Web Search Prompt Caching Audio Input Audio Output

Specifications

Model Keymistral/devstral-medium-latest
ProviderMistral AI
Provider IDmistral
ModeText
Canonical Namedevstral-medium
Context Window256K tokens
Max Output256K tokens

Pricing

TypePer 1K TokensPer 1M Tokens
Input Tokens0.0004000.400
Output Tokens0.00202.00

Benchmarks

Intelligence Index15.1
Coding Index15.9
Math Index4.7
MMLU-Pro0.7
GPQA0.5
HLE0.0
LiveCodeBench0.3
AIME0.1
IFBench0.3
Time to First Token0.43s
SciCode0.3
MATH-5000.7
AIME 20250.0
LCR0.3
TerminalBench Hard0.1
TAU20.2