Pricing and Balance¶
Currency¶
For Nexus users, all billing is in rubles:
- balance top-ups;
- temporary reservation before a request;
- charge after a successful request;
- manual adjustments.
Test Access¶
Test access can be requested from support. Billing is in rubles only.
Estimated Prices¶
The prices below are shown in rubles per 1 million tokens and may change. Check model availability for your key with GET /v1/models.
Prices are derived from IKunCode upstream CNY costs, converted to RUB, and include Nexus FX conversion markup and service markup. The public table is exported from the admin pricing source daily; the Telegram bot displays prices from the same source with a shorter refresh cadence.
Generated Pricing Table¶
Generated by nexus-admin pricing export-docs --update-user-docs from upstream_model_price and pricing_settings. Recommended public docs cadence: once per day.
Claude family¶
| Endpoint | Model | Input RUB / 1M | Cache RUB / 1M | Cache write 5m RUB / 1M | Output RUB / 1M |
|---|---|---|---|---|---|
chat.completions, messages |
claude-haiku-4-5-20251001 |
~5.83 | ~0.58 | ~7.29 | ~29.17 |
chat.completions, messages |
claude-sonnet-4-6 |
~17.50 | ~1.75 | ~21.88 | ~87.50 |
chat.completions, messages |
claude-opus-4-6 |
~29.17 | ~2.92 | ~36.46 | ~145.83 |
chat.completions, messages |
claude-opus-4-7 |
~29.17 | ~2.92 | ~36.46 | ~145.83 |
chat.completions, messages |
claude-opus-4-8 |
~29.17 | ~2.92 | ~36.46 | ~145.83 |
GPT / Codex family¶
| Endpoint | Model | Input RUB / 1M | Cache RUB / 1M | Cache write 5m RUB / 1M | Output RUB / 1M |
|---|---|---|---|---|---|
chat.completions, completions, responses |
codex-auto-review |
~5.83 | ~0.58 | - | ~35.00 |
chat.completions, responses |
gpt-5.4 |
~5.83 | ~0.58 | - | ~35.00 |
chat.completions, responses |
gpt-5.4-mini |
~1.75 | ~0.17 | - | ~10.50 |
chat.completions, responses, responses.compact |
gpt-5.4-openai-compact |
~5.83 | ~0.58 | - | ~35.00 |
chat.completions, responses |
gpt-5.5 |
~11.67 | ~1.17 | - | ~70.00 |
chat.completions, responses |
gpt-5.5-openai-compact |
~11.67 | ~1.17 | - | ~70.00 |
How Balance Works¶
- the balance is tracked at the Nexus profile level;
- multiple API keys on the same profile spend the shared profile balance;
- if there are not enough funds, a new request is rejected;
- after a top-up, the profile can be used again.
Check the current balance with:
curl -H "Authorization: Bearer $NEXUS_API_KEY" \
https://api.nexus-hub.ru/v1/balance
What Happens During a Request¶
- Nexus checks the key, model, and balance.
- It temporarily reserves an amount based on the maximum possible request cost.
- After the request finishes, the reserve is released.
- The actual cost in rubles is charged.
The actual charge is often lower than the reserve because the reserve includes a safety margin.
Manual Review¶
Sometimes the cost cannot be calculated reliably automatically, for example if the response was interrupted or the client received only part of it and the final usage data never arrived.
In that case, the request is sent for manual review. This helps avoid overcharging and prevents the balance from being pushed negative automatically.