Models & Pricing

Models & Pricing

Phase 1 has 34 active models from 12 model providers: 32 Amazon Bedrock models plus GPT-5.6 Terra and Luna through Azure. All use one Nexotao key and one Rupiah balance.

Pricing policy

Nexotao retail rates are a manual snapshot at 100% of each live upstream’s Standard price, converted at Rp 18,000/USD. The reference is AWS Bedrock US Standard for the 32 Bedrock models and Azure/OpenAI Standard for GPT-5.6 Terra and Luna; prices are not synchronized automatically.

The production catalog is the source of truth. Read it before making cost assumptions:

curl -s https://api.nexotao.com/models | jq '.models[] | {
  model, display_name, provider, supports_tools, supports_vision,
  context_window, input_per_million_micro, output_per_million_micro
}'

Agentic models (26)

These passed Nexotao’s stable tool-call round-trip test and report supports_tools: true:

ProviderAPI Model IDs
Anthropicclaude-opus-4-6, claude-opus-4-5, claude-sonnet-4-6, claude-sonnet-4-5, claude-haiku-4-5
Mistralmistral-large-3, devstral-2, magistral-small, ministral-14b, ministral-8b
Amazonnova-micro, nova-lite, nova-pro, nova-2-lite
Metallama-4-maverick-17b
Qwenqwen3-32b, qwen3-coder-30b, qwen3-coder-next
NVIDIAnemotron-super-120b
Z.AIglm-5, glm-4-7
MiniMaxminimax-m2
Moonshotkimi-k2-5
OpenAIgpt-5.6-terra, gpt-5.6-luna, gpt-oss-120b

Good defaults:

  • Claude Code: claude-opus-4-6, claude-sonnet-4-6, or claude-haiku-4-5.
  • Coding agents: qwen3-coder-next, qwen3-coder-30b, or Claude.
  • Fast and economical agentic work: nova-micro.
  • Codex CLI: gpt-5.6-terra or gpt-5.6-luna through /v1/responses.
  • General-purpose agentic work: gpt-oss-120b, glm-4-7, or mistral-large-3.

Chat-only models (8)

These are available for chat and text generation but report supports_tools: false; do not select them for a client that requires tool calling:

deepseek-r1, deepseek-v3-2, ministral-3b, llama-4-scout-17b, llama-3-3-70b, nemotron-nano-30b, gemma-3-27b, gpt-oss-20b.

Live rates

Rates below are fetched in the browser from GET /models. Model IDs are case-sensitive. If this page and the API ever disagree, trust the API.

Loading live pricing…

Every text rate is per 1 million tokens, separately for input and output. The amount billed is based on the production catalog rate at request time. See Billing & Pricing for accounting details.

Any Other ToolBilling & Pricing