Pricing
Inference-first pricing with clear paths for dedicated capacity and GPU Workspaces.
Transparent pricing for every deployment path
Inference first. Dedicated endpoints for enterprise traffic. GPU Workspaces when you need to build before serving.
Monthly inference estimator
Uses model pricing from this repository's catalog. No synthetic cross-provider discount assumptions.
Monthly estimate
Reserved capacity pricing will be published at launch.
For proprietary-routed models, token rates are pass-through from provider list pricing; Brightnode adds routing, latency, and observability layers.
Managed model pricing (per 1M tokens)
Popular live models pulled from the repository model catalog.
| Family | Model | Route | Context | Input | Output | Region | Status |
|---|---|---|---|---|---|---|---|
| Claude | Claude Sonnet 4 | Proprietary models | 200,000 | $3.00 | $15.00 | Singapore, Sydney, Tokyo, Thailand, Malaysia, Jakarta, New Zealand, Seoul, Taiwan, Mumbai | Live |
| Claude | Claude Haiku 4.5 | Proprietary models | 200,000 | $1.00 | $5.00 | Singapore, Jakarta, Malaysia, Thailand, Tokyo, Seoul, Taiwan, Mumbai, Sydney, New Zealand | Live |
| Llama | Llama 3.3 70B Instruct | Brightnode-hosted | 131,072 | $0.22 | $0.50 | Singapore | Live |
| Qwen | Qwen3 32B | Brightnode-hosted | 131,072 | $0.10 | $1.20 | Singapore | Live |
| Deepseek | DeepSeek V3 | Proprietary models | 163,840 | $0.60 | $1.74 | Jakarta, Singapore, Malaysia, Thailand, Tokyo, Seoul, Taiwan, Mumbai, Sydney, New Zealand | Live |
| Mistral | Mistral Nemo | Brightnode-hosted | 131,072 | $0.15 | $0.15 | Singapore | Live |
Bnodes (GPU Workspaces)
Secondary product lane for fine-tuning and eval before deploy-to-inference.
| GPU | vRAM | Price from | Region | Best for |
|---|---|---|---|---|
| T4 | 16GB | $0.50/hr | Singapore | ComfyUI, prototyping, lightweight model work |
| L4 | 24GB | $0.87/hr | Singapore | Embedding pipelines and medium-size inference tests |
| A100 | 80GB | $4.01/hr | Singapore | Fine-tuning, evaluation suites, 70B+ experimentation |
| H100 | 80GB | $14.29/hr | Singapore | Heavy training and high-throughput pre-production validation |
| B200 | 180GB | On request | Singapore | Frontier-scale workloads with reserved capacity |
Dedicated endpoints (coming soon)
Enterprise-grade reserved capacity with regional deployment control.
A100 Endpoint
A100 80GB · Singapore
Dedicated throughput for production chat and agents
Coming soon
H100 Endpoint
H100 80GB · Singapore
High-throughput enterprise inference and heavy traffic
Coming soon
Pay per second. No hourly minimums. No commitments.
Network egress: free within APAC regions
Persistent storage: $0.044/GB/month