Deprecation

To continuously improve service quality, Ant Ling regularly introduces newer model versions with better performance and lower cost, while retiring older versions. This page outlines the deprecation policy, deprecated models, and migration recommendations.

Deprecation Policy

Advance notice: We will notify users via documentation and announcements before a model is officially deprecated.
Service termination: After the deprecation date, API calls using the affected model ID will return errors and no longer be served.
Smooth migration: Each deprecated model comes with a recommended replacement that is no worse in performance and cost than the original.

Deprecated Models

The following models will be retired from the API Console on 2026-06-15 00:00:00 (UTC+8). Please migrate to the recommended replacements.

Model ID	Family	Deprecation Date	Recommended Replacement
`Ling-1T`	Ling 2.0	2026-06-15 00:00:00 (UTC+8)	`Ling-2.6-1T`
`Ring-1T`	Ring 2.0	2026-06-15 00:00:00 (UTC+8)	`Ring-2.6-1T`

Currently Available Models

The following models are in stable support and can be safely used in production.

Model ID	Family	Description
`Ling-2.6-1T`	Ling 2.6	Flagship general model, ideal for complex tasks and long-context scenarios
`Ling-2.6-flash`	Ling 2.6	Cost-efficient model, ideal for high-concurrency online services
`Ring-2.6-1T`	Ring 2.6	Flagship reasoning model, ideal for agentic workflows and complex reasoning
`Ling-2.5-1T`	Ling 2.5	Previous-generation general model, for complex tasks and long-context scenarios
`Ring-2.5-1T`	Ring 2.5	Previous-generation reasoning model, for agentic workflows and complex reasoning

Migration Guide

Ling

Replace the legacy model ID in your code with the corresponding new version:


# Legacy (deprecated)
model="Ling-1T"
 
# New (recommended)
model="Ling-2.6-1T"    # Flagship: complex reasoning, long documents, agentic tasks

Selection guidance:

Previously using Ling-1T or Ling-2.5-1T for everyday chat and content generation → migrate to Ling-2.6-flash, which has fewer active parameters and lower inference cost.
Previously used for complex reasoning, long-context, or agentic tasks → migrate to Ling-2.6-1T, which supports up to 1M token context and delivers significantly improved task completion.

Ring


# Legacy (deprecated)
model="Ring-1T"
 
# New (recommended)
model="Ring-2.6-1T"

Ring-2.6-1T introduces adjustable reasoning effort (high / xhigh) compared to the legacy version, allowing flexible control over reasoning depth and token cost based on task complexity. No other parameter changes are needed after migration.

Third-Party Platforms

When integrating via Zenmux, update the model ID to the new version as well:


# Zenmux
inclusionai/ling-2.6-1t
inclusionai/ling-2.6-flash
inclusionai/ring-2.6-1t

For any questions, refer to the OpenRouter integration guide or Zenmux integration guide.

Was this page helpful?