DeepSeek
DeepSeek V3.2
GPT-4-class quality at sub-cent pricing. Battle-tested predecessor to V4.
ValueFast
$ / 1M input
$0.252
What you pay per million prompt tokens.
$ / 1M output
$0.378
What you pay per million completion tokens.
Blended (70 / 30)
$0.290
Typical chat workload mix.
Ratings & benchmarks
Snapshot 2026-04-28Overall
4.5 / 5
Composite of intelligence + reliability.
Output speed
60 t/s
Output tokens per second under typical load.
Time to first token
1.0 s
Lower is better. Reasoning models naturally stretch this.
Value
4.9 / 5
Intelligence per dollar at typical mix.
Context window
131,072 tokens