L LLM Cloud Hub
Vendor comparison

Mistral vs Perplexity

Every Mistral and Perplexity LLM model side by side: pricing per million tokens, context windows, and capabilities. Refreshed nightly from upstream.

Mistral

24 models
Model Context In $/1M Out $/1M
Codestral 2508
tools, json_mode
256k 0.3000 0.9000
Devstral 2 2512
tools, json_mode
262k 0.4000 2.0000
Devstral Medium
tools, json_mode
131k 0.4000 2.0000
Devstral Small 1.1
tools, json_mode
131k 0.1000 0.3000
Ministral 3 14B 2512
vision, tools, json_mode
262k 0.2000 0.2000
Ministral 3 3B 2512
vision, tools, json_mode
131k 0.1000 0.1000
Ministral 3 8B 2512
vision, tools, json_mode
262k 0.1500 0.1500
Mistral 7B Instruct v0.1 3k 0.1100 0.1900
Mistral Large
tools, json_mode
128k 2.0000 6.0000
Mistral Large 2407
tools, json_mode
131k 2.0000 6.0000
Mistral Large 2411
tools, json_mode
131k 2.0000 6.0000
Mistral Large 3 2512
vision, tools, json_mode
262k 0.5000 1.5000
Mistral Medium 3
vision, tools, json_mode
131k 0.4000 2.0000
Mistral Medium 3.1
vision, tools, json_mode
131k 0.4000 2.0000
Mistral Medium 3.5
vision, tools, json_mode
262k 1.5000 7.5000
Mistral Nemo
tools, json_mode
131k 0.0200 0.0300
Mistral Small 3
json_mode
33k 0.0500 0.0800
Mistral Small 3.1 24B
vision
128k 0.3500 0.5600
Mistral Small 3.2 24B
vision, tools, json_mode
128k 0.0750 0.2000
Mistral Small 4
vision, tools, json_mode
262k 0.1500 0.6000
Mixtral 8x22B Instruct
tools, json_mode
66k 2.0000 6.0000
Pixtral Large 2411
vision, tools, json_mode
131k 2.0000 6.0000
Saba
tools, json_mode
33k 0.2000 0.6000
Voxtral Small 24B 2507
tools, json_mode
32k 0.1000 0.3000

Perplexity

5 models
Model Context In $/1M Out $/1M
Sonar
vision
127k 1.0000 1.0000
Sonar Deep Research 128k 2.0000 8.0000
Sonar Pro
vision
200k 3.0000 15.0000
Sonar Pro Search
vision
200k 3.0000 15.0000
Sonar Reasoning Pro
vision
128k 2.0000 8.0000

Keyboard shortcuts

?
Show this overlay
/
Focus the first form field
g h
Go to / (home)
g b
Go to /best-llm-for
g c
Go to /cost
g s
Go to /self-hosted
g x
Go to /compliance
Esc
Close any overlay

Inspired by Linear and GitHub conventions. The two-key sequences (g then h) work within ~1 second.