HAIL.SO / DISPATCH · 2026-07-25 · COMPARE
FILE: COMPARE · 6 of 6 slots

side by side.

← all costsclear
01

Large language models

4 models
DeepSeek
DeepSeek V3
×
Alibaba
Qwen3.7-Max
×
Mistral
Ministral 3 14B
×
Perplexity
Sonar
×
Output $/MTok$0.28$7.50$0.20$1.00
Input $/MTok$0.14$2.50$0.20$1.00
Cached input $/MTok$0.0028
Context window1,048,5761,048,576262,144128,000
Output cap384,00065,5368,1928,000
Modalitiesin: text / out: textin: text / out: textin: text, image / out: textin: text / out: text
Tool use
Structured output
FamilyDeepSeek V3Qwen 3.7 MaxMinistralSonar
Knowledge cutoff
Verified2026-07-192026-07-192026-07-192026-07-19
Sourceprovider page ↗provider page ↗provider page ↗provider page ↗
03

Text to speech

2 models
LMNT
LMNT Blizzard
×
Deepgram
Deepgram Aura 2
×
$/1M chars$50.00$30.00
Voice qualityneuralneural
Voice cloning✓ included
Voice count
Languages31+en, es, de, fr, nl, it, ja
SSML support
TTFB
Output formatswav, mp3, linear16, mulaw, alaw, opus
Verified2026-07-192026-07-19
Sourceprovider page ↗provider page ↗