Budget · Under $1/M
LLMs under $1 per million tokens
Every model priced below $1 per 1M input tokens. Ranked by benchmark score within budget.
Models40
Top quality78.4
Max price$1.00/M
What this page is
This page lists every model with input priced below $1 per million tokens, ranked by benchmark quality. This is the volume tier · most production traffic for chat, classification, extraction, and bulk content runs on sub-$1 models. The quality variance within this budget is huge, so the ranking matters.
Ranked by quality within budget
Input under $1/M, highest benchmark score first.
Top 3 best-quality sub-$1 models
Best quality under $1
Qwen3.5 397B A17B
input
$0.39/M
output
$2.34/M
Qwen3.5 397B A17B is the highest-scoring model with input priced below $1/M. Excellent for bulk, high-volume workloads.
Runner up
DeepSeek V3.2 Speciale
input
$0.40/M
output
$1.20/M
DeepSeek V3.2 Speciale is the highest-scoring model with input priced below $1/M. Excellent for bulk, high-volume workloads.
Third
Step 3.5 Flash
input
$0.10/M
output
$0.30/M
Step 3.5 Flash is the highest-scoring model with input priced below $1/M. Excellent for bulk, high-volume workloads.
The price gap · cheapest vs most expensive
Cheapest
Qwen3.5 397B A17B
$0.39/M
$ per 1M input tokens
Why the gap
At the sub-$1 tier, the "most expensive" is still cheap. The tradeoff is raw benchmark quality vs provider reliability. Open-source models dominate the cheap end.
Most expensive
Qwen3 Max
$0.78/M
$ per 1M input tokens
Frequently asked questions
Most open-source models (DeepSeek V3, Qwen3.5, GLM-4.6, Llama 3.3) plus small proprietary models (Gemini Flash, GPT-4o-mini, Claude Haiku) on various providers. The cheap end clusters around $0.05 to $0.30 per million.