Models & Providers

Tackle is built on laravel/ai and is provider-agnostic like it: any provider laravel/ai supports with tool calling works. Anthropic (Claude) is the default.

Changing the model or provider

The quickest way is via .env:

AI_CODE_PROVIDER=openai
AI_CODE_MODEL=gpt-4o

The provider name must match a key in config/ai.php. Any provider supported by laravel/ai (Anthropic, OpenAI, Gemini, Groq, Ollama, etc.) works as long as it supports tool calling.

When switching models, also set AI_CODE_PRICE_INPUT / AI_CODE_PRICE_OUTPUT to the model's per-million-token rates — the budget cap is estimated from token counts, and the defaults assume Claude Sonnet pricing. For local models set both to 0:

AI_CODE_PROVIDER=ollama
AI_CODE_MODEL=qwen3-coder
AI_CODE_PRICE_INPUT=0
AI_CODE_PRICE_OUTPUT=0

Subscription-based providers

Providers don't have to be API-key services: the Tackle Codex add-on registers a codex driver that runs the agents on OpenAI's Codex models via your ChatGPT subscription, with usage recorded as $0 against the session budget, and Tackle Grok adds a grok driver for xAI's Grok models. Both are working examples of packaging a custom laravel/ai driver for Tackle.

The model catalog

Budget rates resolve automatically from Tackle's built-in model catalog (Anthropic, plus common OpenAI, Gemini, and Grok models — run /model in ai:code to see them with their rates). For models the catalog doesn't know, pin rates explicitly so budget enforcement stays meaningful.

You can also teach the catalog new models (or correct a stale built-in rate) in config/tackle.php under pricing.models. Non-Anthropic built-in rates are best-effort snapshots — verify against your provider's pricing page when the budget matters.

Per-session and mid-session switching

Switch models per session with ai:code --model=... [--provider=...] (also on ai:run), or mid-session with the /model slash command — no name shows a picker listing known models with their per-MTok rates.

How it's wired

Internally, Tackle injects provider and model values via two custom Laravel contextual attributes — #[AiProvider] and #[AiModel] — so any agent you write by extending DefaultCodingAgent inherits these config values automatically through the container.

Agent quality tracks the model

The coding and review agents lean heavily on tool calling, so weaker models produce weaker results. The plumbing is neutral; our recommendation is Claude.