Kilo Gateway: Add GLM 5.1, Remove MiniMax M2.5 Free

[Ardakilic]

Kilo Gateway: Add GLM 5.1, Remove MiniMax M2.5 Free

Summary

Two model changes for the Kilo Gateway provider:

1. Add z-ai/glm-5.1 — GLM 5.1 is available on the Kilo API and already listed by other providers on models.dev (OpenRouter, Zenmux), but was missing from the Kilo entry.
2. Remove minimax/minimax-m2.5:free — the free tier of MiniMax M2.5 has been removed from the Kilo API and the TOML file is deleted accordingly.

Changes

• New file: providers/kilo/models/z-ai/glm-5.1.toml
• Deleted file: providers/kilo/models/minimax/minimax-m2.5:free.toml

Model Details

Field	Value
Name	Z.ai: GLM 5.1
Provider	Kilo Gateway
Model ID	z-ai/glm-5.1
Release date	2026-03-27
Context window	202,752 tokens
Max output	131,072 tokens
Input cost	$1.26 / 1M tokens
Output cost	$3.96 / 1M tokens
Reasoning	Yes
Tool calling	Yes
Temperature	Yes
Open weights	Yes
Modalities	Text → Text

Field Rationale

• Naming: Follows the "Z.ai: <Model>" convention used by all other Kilo z-ai models (GLM 5, GLM 4.7, GLM 4.5, etc.)
• Costs: Derived from Kilo API pricing data — $0.00000126/token input → $1.26/M, $0.00000396/token output → $3.96/M
• Context / output limits: context = 202752 matches the API's context_length; output = 131072 matches GLM 5 on Kilo (the API returns null for max output, so the GLM 5 value is used as a conservative ceiling)
• Capabilities: reasoning, tool_call, and temperature are all confirmed by the Kilo API's supported params list for this model
• open_weights = true: Consistent with GLM 5.1 on OpenRouter and all other Kilo z-ai models
• attachment = false: Text-only input/output; no image or file attachment support listed by the API

Fields Intentionally Omitted

• family: No existing Kilo z-ai model uses this field
• structured_output: Not used by any Kilo z-ai model (even though the API supports it)
• [interleaved]: Not used by any Kilo z-ai model
• cache_read / cache_write: Not present in Kilo API pricing data for this model

Validation

bun validate passes with no errors. The model appears correctly in the validated output as:

"z-ai/glm-5.1": {
  "id": "z-ai/glm-5.1",
  "name": "Z.ai: GLM 5.1",
  "attachment": false,
  "reasoning": true,
  "tool_call": true,
  "temperature": true,
  ...
}