What does the free tier include?

A daily usage allowance on the affordable and standard model bands, no card required; it resets every day. Premium models ask for an upgrade. Local inference adds 1000 free local requests per rolling 30 days on your own hardware.

Are free chats used for training?

No. idapt never trains on your data, free tier included.

What happens when the daily allowance runs out?

The app tells you before you send, and the allowance resets the next day. Local models keep working through your own machine, and plans from $25/month raise the included usage.

All roundups

Best of

The models that cost nothing to run.

The two free lanes, compared: catalog models covered by the free daily allowance, and local models running on your own machine.

Last reviewed July 16, 2026

Try idapt free Browse all models

idapt has two lanes where a model costs nothing: the free daily usage allowance, which covers the catalog's free-lane models with no card required, and local inference, where open models run through Ollama on your own machine. The picks below are the strongest models in those lanes right now; premium models are out of scope here because they sit behind an upgrade.

How we pickedEvery pick is verifiably free on idapt in one of two lanes: it sits in the free-lane catalog covered by the daily usage allowance, or it runs locally through Ollama with no per-token price. Membership was checked against the live registry at review; specs beside each pick render from the same data.

The picks

1Gemma 4 31BThe free-lane default.

$0.12 in / $0.35 out per M tokens · 262K context

The model the free tier leans on: capable general answers, image input, and open weights that also run locally on a well-equipped workstation.

Best for: The everyday default at zero cost.

2DeepSeek V4 FlashLong context in the free lane.

$0.09 in / $0.18 out per M tokens · 1M context

A million-token context window on a free-lane model: whole reports and codebases fit in one prompt without touching a balance.

Best for: Long documents without spending.

3Qwen3.5 FlashFree vision at speed.

$0.065 in / $0.26 out per M tokens · 1M context

Image input and a million-token window in the free lane: screenshots, charts, and scans get read without an upgrade.

Best for: Image-heavy work at zero cost.

4Qwen3 235B A22BThe biggest free brain.

$0.09 in / $0.1 out per M tokens · 262K context

A 235B-parameter open MoE (22B active) in the free lane; the pick when a free-tier question deserves a heavyweight.

Best for: Harder questions while staying free.

5Qwen3 8BThe local lane on modest hardware.

$0.05 in / $0.4 out per M tokens · 41K context

Free both ways: in the catalog's free lane and locally via Ollama, where its q4 build recommends just 8 GB of RAM.

Best for: Private, offline, zero-cost chat.

Frequently asked

Run them in one workspace

Multi-Model Chat

GPT, Claude, Gemini, Grok, Llama: pick mid-conversation, cheapest-first.

Local AI

Run open models on your own hardware with Ollama: private, free, no rate limits.

Keep comparing

Frontier answers at commodity prices.Serious models, your hardware.Open weights at the frontier.

Judge them on your own prompt

Every pick on this page runs in one idapt chat. Start free with a daily allowance, no credit card.

Try idapt free