HFAI Cloud · model picker
Choose by constraint.
Not by opinion.
Every row below comes from the catalogue the API serves: five numbers and no ranking.
Five questions, five answers.
Each block is recomputed
from the live catalogue.
This page asks five questions of the roster the API publishes: how much context a row holds, which rows cost least to read from and write to, and which row the free tier covers. Every answer is a number or a list, and both move when the catalogue moves.
At least 256K of context
5 of 11 rows fit: DeepSeek V4 Pro, Nemotron 3 Ultra, Kimi K2.6, Neira H4 and GLM-5.3 Abliterated.
Their windows run from 256,000 to 1,000,000 tokens.
A full 1M of context
2 of 11 rows reach 1,000,000 tokens: DeepSeek V4 Pro and GLM-5.3 Abliterated.
Every row in this set also clears 256,000 tokens.
Cheapest input
The lowest input price is $0.07 per 1M tokens, on GPT-OSS 20B.
The next input price above it is $0.10, on Nemotron 3 Ultra.
Cheapest output
The lowest output price is $0.30 per 1M tokens, on GPT-OSS 20B.
The next output price above it is $0.40, on Gemma 4.
The only free row
GPT-OSS 120B is the only row the free tier covers: $0.15 per 1M in and $0.60 per 1M out.
The other 10 rows are marked not free.
One table, 11 rows.
Every name links its own
model page.
| Model | Context window | Price in / 1M | Price out / 1M | Free row |
|---|---|---|---|---|
| DeepSeek V4.1 Flash | 128,000 tokens (128K) | $0.30 | $1.20 | No |
| GPT-OSS 120B | 128,000 tokens (128K) | $0.15 | $0.60 | Yes |
| GPT-OSS 20B | 128,000 tokens (128K) | $0.07 | $0.30 | No |
| Gemma 4 | 128,000 tokens (128K) | $0.14 | $0.40 | No |
| GLM 5.3 Flash | 128,000 tokens (128K) | $0.15 | $0.50 | No |
| MiniMax M3 | 128,000 tokens (128K) | $0.60 | $2.40 | No |
| DeepSeek V4 Pro | 1,000,000 tokens (1M) | $1.32 | $3.96 | No |
| Nemotron 3 Ultra | 256,000 tokens (256K) | $0.10 | $3.00 | No |
| Kimi K2.6 | 256,000 tokens (256K) | $0.95 | $4.00 | No |
| Neira H4 | 262,144 tokens (262144) | $1.20 | $3.60 | No |
| GLM-5.3 Abliterated | 1,000,000 tokens (1M) | $3.60 | $6.00 | No |
Read down a column to compare one value across every model, or follow a name to its own page for the full record. The five blocks above reduce the same columns to their ends: the widest windows, the lowest prices and the free flag.
From a number to a request.
The pricing page explains how the prepaid balance works and how credits are bought. The cloud page carries the sign-in and key steps. Client setup lives under for developers, and the request shape is documented in the API reference.