Supported AI models
LLM models available in the bitERP chat (app.biterp.ai), with status and end-of-support dates.
This page describes the app.biterp.ai chat
The catalog below applies to the model picker in the product chat (app.biterp.ai).
It does not apply to the MCP Server (mcp.biterp.ai), where the
model is chosen by your own AI client, nor to the Integrations API,
which does not use an LLM.
Model status
| Status | Meaning |
|---|---|
| Available | Selectable in the chat model picker |
| Internal use | Not selectable; the platform uses it in auxiliary steps of assistance (consumes tenant credits) |
| End of support announced | Still selectable, with an exit date published in the "End of support" column |
| Discontinued | Removed from the catalog; anyone who had it selected has already been migrated to the substitute |
Available models
Models you can choose in the chat picker.
| Model | Provider | Status | Intelligence | Relative cost | End of support | Substitute |
|---|---|---|---|---|---|---|
| GPT-6.1 Sol | OpenAI | Available | 5 | 3 | — | — |
| GPT-5.6 Terra | OpenAI | Available | 4 | 4 | — | — |
| GPT-6 Luna | OpenAI | Available | 3 | 1 | — | — |
| Claude Opus 5.5 | Anthropic | Available | 5 | 5 | — | — |
| Claude Sonnet 5.5 | Anthropic | Available | 5 | 3 | — | — |
| Grok 4.6 | xAI | Available | 4 | 2 | — | — |
Both scores range from 1 to 5 and are the same scales shown in the chat picker. Intelligence reflects the general capability index published by Artificial Analysis for each model; relative cost reflects the price band (1 = most economical, 5 = most expensive). Neither is a per-token price.
How to choose a model
In the app.biterp.ai chat, use the model picker to choose which LLM the main agent runs on. The choice applies to that conversation and consumes tenant credits based on usage.
In general:
- Models with higher intelligence handle complex, multi-step tasks and dense business rules better.
- Models with a lower relative cost prioritize cost and latency, with capacity suited to day-to-day operations.
- The two scores are independent: a model can have high intelligence and low cost. When in doubt, start with the highest intelligence within the cost you are willing to pay.
There is no model configuration in the Integrations API or MCP — see What this page does not cover.
Internal-use models
These models do not appear in the picker. The platform uses them in auxiliary steps of assistance (for example, specialists that support the main agent). Their usage is still debited from tenant credits.
| Model | Provider | Status | Intelligence | Relative cost | End of support | Substitute |
|---|---|---|---|---|---|---|
| Claude Haiku 4.5 | Anthropic | Internal use | 1 | 2 | — | — |
Discontinuation policy
When a provider ships a new version (or retires the previous one), bitERP updates the chat catalog. The change does not break your selection:
- The saved selection migrates automatically to the substitute in the same family (same provider and capability band). When the provider has no model left in the chat, it migrates to a model from another provider in the same band: that was the case of DeepSeek V4.1 Flash, replaced by GPT-6 Luna.
- You do not need to re-select the model after the change.
- When an exit date is announced, it appears in the End of support column as
YYYY-MM-DD. With no announced date, the column shows—.
End-of-support dates depend on the provider and the catalog we consume. This page describes how we announce and migrate — it does not promise a fixed minimum notice window.
Discontinued models
History of models that left the picker. Anyone who had them selected was migrated to the substitute below.
| Discontinued model | Current substitute |
|---|---|
| GPT-5.6 Sol | GPT-6.1 Sol |
| GPT-5.6 Luna | GPT-6 Luna |
| Claude Opus 5 | Claude Opus 5.5 |
| Claude Sonnet 5 | Claude Sonnet 5.5 |
| DeepSeek V4.1 Flash | GPT-6 Luna |
| GPT-5.5 | GPT-6.1 Sol |
| GPT-5.4 | GPT-5.6 Terra |
| GPT-5.4 mini | GPT-6 Luna |
| GPT-5.2 | GPT-6.1 Sol |
| GPT-5.2 chat | GPT-6.1 Sol |
| GPT-5.1 | GPT-5.6 Terra |
| GPT-5 mini | GPT-6 Luna |
| Claude Opus 4.7 | Claude Opus 5.5 |
| Claude Opus 4.6 Fast | Claude Sonnet 5.5 |
| Claude Sonnet 4.6 | Claude Sonnet 5.5 |
| Claude Sonnet 4.5 | Claude Opus 5.5 |
What this page does not cover
- MCP Server (
mcp.biterp.ai): the server only exposes tools via the MCP protocol. The model (and token cost) is chosen by the user's AI client — Claude Desktop, Cursor, ChatGPT, and similar. - Integrations API: REST API for external systems. It does not involve an LLM.