Curated Models

Browse our curated selection of popular and recommended AI models for getting started with OpenClaw. While hundreds of models are available via the API, these represent the best balance of capability, speed, and cost for general use.

Tag:

All Cheap Starter Super cheap

Model Performance vs. Blended Cost

Model Name ↑	Weekly Usage ↕	$/M Tokens ↕	Release Date ↕	Agent ↕	Value ↕
DeepSeek: DeepSeek V3.2	1,200	$0.28	2025-12-01	77	98
Tag-Cheap This has been a workhorse with many reports of it being a great model for the low cost. It is a distilled model, so it can disappoint in odd ways.
Google: Gemma 4 31B	7	$0.19	2026-04-02	73	18
TAG-Cheap
OpenAI: gpt-oss-120b	448	$0.07	2025-08-05	72	8
TAG-Cheap Much older model that has continued to perform okay at a very low price point. Great for exploration without spending money.
Qwen: Qwen3.5-9B	-	$0.07	2026-03-10	84	392
TAG-Cheap
Z.ai: GLM 4.7 Flash	-	$0.13	2026-01-19	78	128
TAG-cheap
xAI: Grok 4.1 Fast	828	$0.26	2025-11-19	64	-
TAG-Starter TAG-Cheap Has been known as a frontier model, but XAI has been slow to update, so this model does not have the capabilities of new models.

* Weekly Usage is the total tokens processed (in Billions) across the OpenRouter network.
* Cost/M Token is a blended estimate per 1 Million tokens, calculated as (Input * 4 + Output) / 5.
* Agent represents the average of agentic capabilities benchmarks in the Artificial Analysis Intelligence Index (GDPval-AA, 𝜏²-Bench Telecom). See https://artificialanalysis.ai/.
* Value is calculated as (Agent - 70)² / Max(Cost, 0.5) for Agent scores > 70.