How are efficient models ranked?

Each model is scored on weighted, normalized metrics (pricing, capability, and benchmark data), then adjusted for data confidence and freshness. Sources are listed on every model page.

How fresh is this data?

Data is verified against named sources and flagged when older than 45 days. The last verification date is shown on each page.

Efficient Models

Low-cost, low-latency models for high-volume workloads.

5 models · Updated Jul 12, 2026

#		provider	model category
1	DeepSeek V4 Flash	DeepSeek	—	62.8
7	Gemini 3.5 Flash	Google	—	29.3
8▲1	Claude Haiku 4.5 (latest)	Anthropic	—	29.1
15▲1	Llama 4 Maverick 17B Instruct	Meta	—	18.7
16▲1	o4-mini	OpenAI	—	18.1

Ranking combines weighted, normalized metrics with confidence and freshness adjustments. Missing data lowers a score rather than being ignored.

Popular comparisons

DeepSeek V4 Flash vs Gemini 3.5 Flash
DeepSeek V4 Flash vs Claude Haiku 4.5 (latest)
Gemini 3.5 Flash vs Claude Haiku 4.5 (latest)

Related model categories

Flagship Models
Coding Models

Frequently asked questions

How are efficient models ranked?: Each model is scored on weighted, normalized metrics (pricing, capability, and benchmark data), then adjusted for data confidence and freshness. Sources are listed on every model page.
How fresh is this data?: Data is verified against named sources and flagged when older than 45 days. The last verification date is shown on each page.