HAIL.SO / DISPATCH · 2026-07-22 · COMPARE
FILE: COMPARE · 6 of 6 slots

side by side.

← all costsclear
01

Large language models

4 models
Perplexity
Sonar
×
Alibaba
Qwen3-Max
×
OpenAI
GPT-5.4
×
Mistral
Ministral 3 3B
×
Output $/MTok$1.00$6.00$15.00$0.10
Input $/MTok$1.00$1.20$2.50$0.10
Cached input $/MTok$0.2500
Context window128,000262,1441,050,000262,144
Output cap8,00032,768128,0008,192
Modalitiesin: text / out: textin: text / out: textin: text, image / out: textin: text, image / out: text
Tool use
Structured output
FamilySonarQwen 3 MaxGPT-5.4Ministral
Knowledge cutoff2025-06-302025-08-31
Verified2026-07-192026-07-192026-07-192026-07-19
Sourceprovider page ↗provider page ↗provider page ↗provider page ↗
02

Speech to text

1 model
Rev.ai
Rev.ai Reverb
×
$/min$0.003300
$/min batch
Streaming
Realtime
Languagesen
Diarizationincluded
TTFW
Verified2026-07-19
Sourceprovider page ↗
03

Text to speech

1 model
LMNT
LMNT Blizzard
×
$/1M chars$50.00
Voice qualityneural
Voice cloning✓ included
Voice count
Languages31+
SSML support
TTFB
Output formats
Verified2026-07-19
Sourceprovider page ↗