Skip to contentQloudOpen models
Console login

Open model access

Open models. Three tiers.

Open models for asynchronous agents and workflows. Compare Realtime, Async and Batch token rates, with cost per completed job in view.

REALTIME BASE URLOpenAI-compatible
https://api.qloud.sh/v1
client = OpenAI(
    base_url="https://api.qloud.sh/v1",
    api_key=QLOUD_API_KEY,
)

client.chat.completions.create(
    model="DeepSeek-V4-Flash-0731",
    messages=messages,
)

OpenAI-compatible chat completions. Async and Batch on the same models.

Three tiers. One workload.

Batch leads on price when work can wait.

Batch inference

$1.20per completed job

$2.80 saved vs Realtime · 70% lower

For latency-tolerant agents and workflows with room to wait.

Input
$0.27
Output
$0.93
Cached input
$0.078

Per 1M tokens, GLM-5.3 list rates.

Batch: $1.20 per completed job.

Compare your job cost

Open weights, tuned infrastructure.

Models are added as they ship. Quantized variants are labelled, never swapped in silently for the checkpoint you asked for.

GLM-5.3

Always reasons1.3M context

Flagship Z.ai mixture-of-experts for coding and long-horizon agent work

GLM-5.3-Flash

Always reasons1.3M context

The fast GLM-5.3 tier, for high-volume agents and interactive work

DeepSeek-V4-Pro

Reasoning1M context

Frontier open reasoning and tool calling for hard, low-volume work

DeepSeek-V4-Flash-0731

Low latency1.3M context

The GA revision of V4 Flash: sparse MoE for coding, reasoning and agents

DeepSeek-V4.1-Flash

Low latency1M context

The V4.1 Flash line: a newer revision than 0731, at a higher rate

Muse-Spark-Open

Always reasons1M context

Meta's open-weight coding model, tuned for agentic workflows and tool calling

Bring the workload.
Keep the weights open.

See how Batch works