Star OS
Models and pricing

Text generation models and pricing

The Qwen, GLM, and DeepSeek text generation models currently available through Star OS.

Star OS currently supports text generation models from Qwen, GLM, and DeepSeek. Use the exact Model ID below as the model value in your request.

How pricing works

Prices are in USD per 1 million tokens and are calculated separately for input and output tokens:

request cost = (input tokens × input rate) + (output tokens × output rate)

The rates below are the current Star OS standard real-time prices. They exclude promotions, batch discounts, and context-cache discounts. For a live list of models enabled for your project, open Models. Your Star OS Console and billing records are the source of truth for charges.

Qwen

Model IDBest forInputOutputNotes
qwen3.7-maxComplex reasoning and demanding production workloads$2.50$7.50Flat rate to 1M input tokens
qwen3.7-plusGeneral-purpose reasoning and agent workflows$0.40$1.60$1.20 / $4.80 above 256K input tokens
qwen3.6-plusBalanced quality for production applications$0.50$3.00$2.00 / $6.00 above 256K input tokens
qwen-plusCost-conscious general text generation$0.40$1.20$1.20 / $3.60 above 256K input tokens
qwen3.6-flashFast, economical responses$0.25$1.50$1.00 / $4.00 above 256K input tokens
qwen3.5-flashHigh-volume, low-latency text generation$0.10$0.40Flat rate to 1M input tokens
qwen-flashSimple, high-volume text generation$0.05$0.40$0.25 / $2.00 above 256K input tokens
qwen3-coder-nextCode generation and software-engineering tasks$0.30$1.50$0.50 / $2.50 above 32K; $0.80 / $4.00 above 128K

GLM

Model IDBest forInputOutputNotes
glm-5.2General text generation and reasoning$1.40$4.40Flat rate; non-thinking and thinking modes
glm-5.2-fast-previewLatency-sensitive generation$2.80$8.80Preview model; non-thinking and thinking modes
glm-5.1General text generation and reasoning$1.40$4.40Rate applies up to 200K input tokens

DeepSeek

Model IDBest forInputOutputNotes
deepseek-v4-proComplex reasoning, coding, and general workloads$2.40$4.80Context-cache discount may apply
deepseek-v4-flashFast, cost-efficient reasoning and text generation$0.20$0.40Context-cache discount may apply
deepseek-v3.2Balanced text generation and reasoning$0.57$1.71Context-cache discount may apply

Selecting a model

Start with a model that matches your workload, then set its Model ID in the request. For example:

{
  "model": "qwen3.5-flash",
  "messages": [{ "role": "user", "content": "Summarise this report." }]
}

If a model is not marked Available in the Console, do not send requests to its ID. Model availability, model profiles, usage, and billing are all scoped to your project.

For throughput limits, see Rate limits.

On this page