Glue — Models & Providers

◇ Models

Any Model.
Any Provider.

Glue doesn't lock you in. Use the best model for the job — swap with a single flag, no restart needed.

■ Supported Providers

First-Class Support

A

Anthropic

Claude 4 Opus, Claude 4 Sonnet, Claude 4 Haiku. Best-in-class for coding tasks. Default provider.

Default Streaming

O

OpenAI

GPT-5, GPT-4.1, o3, o4-mini. Full tool-use support with streaming responses.

Streaming Reasoning

G

Google

Gemini 3 Pro, Gemini 3 Flash. Excellent for large context windows and multi-file analysis.

Streaming 1M Context

◆

OpenAI-Compatible

Any endpoint that speaks the OpenAI API format. Ollama, vLLM, Together, Groq, and more.

Local Self-Hosted

▸ Usage

Switching Models

▸ Model Selection

# Use default (claude-4-sonnet) ❯ glue # Choose a specific model ❯ glue -m gpt-5 ❯ glue -m claude-4-opus ❯ glue -m gemini-3-pro # Local model via Ollama ❯ glue -m ollama/deepseek-r1 # Custom endpoint ❯ glue -m custom --api-url http://localhost:8080

Per-Task Models

Different tasks need different models. Use Opus for complex architecture decisions, Sonnet for everyday coding, Haiku for quick fixes.

Hot Swapping

Switch models mid-session with the /model command. Your conversation context carries over — no restart needed.

Cost Awareness

Glue tracks token usage per session and shows real-time cost estimates in the status bar. Stay in budget.

△ Comparison

Pick the Right Tool

Model	Best For	Speed	Context
Claude 4 Opus	Complex refactors, architecture	Moderate	200K
Claude 4 Sonnet	General coding, best balance	Fast	200K
GPT-5	Reasoning-heavy tasks	Fast	128K
Gemini 3 Pro	Large codebase analysis	Fast	1M
Local (Ollama)	Offline, private, experimentation	Varies	Varies

Any Model.Any Provider.