Search
Browse by category

AI Model Categories

Curated shortlists by use case — coding, reasoning, cheap, open-weight, and more.

10Categories 170Models ranked 41Labs

Overall

The leading models in LMRank overall, ordered by leaderboard rank across the public benchmark mix.

113 models

Coding

The strongest models for code generation, refactoring, debugging, and multi-file edits, based on public coding benchmarks and LMRank ranking.

3 models

Agentic Coding Model

The models we recommend for long-running, agentic coding tasks — multi-step planning, tool use, and code edits.

3 models

Open-Weight

The best open-weight models available on OpenRouter — usable for self-hosting, fine-tuning, and local evaluation.

3 models

Cheap Model

Models that pair strong capability with low listed input pricing. Pricing is compared at list price per million input tokens.

3 models

Multimodal

The strongest multimodal models — those that can take image, audio, or video in alongside text.

3 models

Long Context

Models that support the largest context windows, ideal for long documents, codebases, and multi-turn workflows.

3 models

Reasoning

The best models for multi-step reasoning, math, and chain-of-thought problem solving.

3 models

Fast Model

The lowest-latency variants from each provider — Flash, Turbo, Mini, and Instant models — for interactive apps.

3 models

Local Model

Open-weight models with the smallest parameter counts that still run on consumer hardware.

3 models

AI model category FAQs

How are the categories on LMRank ranked?
Each category orders models by their leaderboard position across the public benchmarks that matter for that job. A model that leads Coding can sit mid-pack in Long Context, because every category is ranked independently against its own criteria.
How often do category rankings update?
Rankings refresh as new models ship and our benchmark suite grows, so each category reflects current measured performance rather than a fixed editorial list. When a new release breaks into the top tier, the category page updates without manual curation.
Can one model appear in more than one category?
Yes. Strong all-rounders show up across several categories, while specialists appear only where they lead. Every category is evaluated on its own, so placement in one has no bearing on another.
Which categories does LMRank track?
Overall, Coding, Reasoning, Agentic Coding, Cheap, Fast, Long Context, Multimodal, Open-Weight, and Local - one focused list per use case, so you can find the right model for the work you actually have.