LLM Reference

LLM Reference is a transformative decision-support directory engineered for developers, architects, and technology leaders who need to navigate the explosive and fast-moving landscape of large language models. It is the definitive, game-changing resource that eliminates the chaos of scattered information, providing a single, trustworthy platform to compare over 1,800 models from 140 providers and 247 research labs. The core value proposition is simple yet powerful: stop wasting precious hours hunting through fragmented sources and start shipping with unshakeable confidence. Whether you are building a next-generation coding assistant, an autonomous agentic workflow, a powerful writing tool, or a complex research pipeline, LLM Reference unlocks your potential by giving you the data you need to make the right choice instantly. The platform tracks over 1,700 models with data refreshed weekly, capturing new releases, verified price changes, and benchmark updates as they happen. Its intuitive interface is designed for fast triage, allowing you to quickly identify the optimal model for your specific job, determine the most cost-effective provider, and get back to building. With a dynamic Pulse feed that highlights exactly what changed this week, including new models, price cuts, and benchmark refreshes, LLM Reference keeps you informed without the noise. It is a Data Advantage project, updated daily, making it an essential, transformative resource for anyone who must stay current with the exploding LLM ecosystem and unlock the full potential of AI.

Comprehensive Model Directory

Access a transformative, searchable directory of over 1,800 language models from 140 providers and 247 labs. This game-changing feature allows you to filter by task, such as coding, agents, writing, research, image generation, or video creation, and instantly see every eligible model. Each entry includes critical data like benchmark scores, pricing per million tokens, and direct links to providers, enabling you to make informed decisions without leaving the platform.

Editors' Picks and Curated Boards

Unlock the potential of expert curation with LLM Reference's Editors' Picks for specific tasks. This transformative feature provides a shortlist of the best models for coding, agents, writing, research, image, and video, validated by real-world performance. Each pick includes a detailed rationale, such as "Claude Fable 5" for coding due to its 80.3% SWE-bench Pro score, allowing you to bypass analysis paralysis and start with a proven, game-changing solution.

Live Pulse and Changelog

Stay ahead of the curve with the game-changing Pulse feed, which highlights what changed in the model market this week. This transformative feature tracks new models (e.g., 177 added this week), verified price cuts (53 this week), and benchmark refreshes (368 this week). The Changelog provides a historical record, ensuring you never miss a critical update that could unlock better performance or lower costs for your projects.

Side-by-Side Model Comparison

Make transformative decisions with the ability to compare two models side-by-side. This feature allows you to pit leading contenders like "Claude Fable 5" against "GPT-5.5" or "Claude Opus 4.8" against "Claude Opus 4.7" on key metrics. You can instantly see differences in benchmark performance, pricing, and context windows, unlocking the potential to choose the absolute best model for your specific requirements without any guesswork.

Selecting a Production Coding Model

A development team building an agentic coding assistant can use LLM Reference to instantly identify the best model for their task. By navigating to the "Coding" section under Editors' Picks, they see that "Claude Fable 5" is the top choice, with an 80.3% SWE-bench Pro score. They can then compare its pricing against other eligible models, ensuring they ship a high-performance solution that stays within budget.

Optimizing Cost for High-Volume API Calls

A startup that processes millions of requests per month needs to find the cheapest frontier output. Using LLM Reference's "Best" and "Frontier pricing" signals, they can quickly identify that "Hunyuan HY3 Preview" via Tencent Cloud TI Platform offers the lowest cost at $0.260 per 1M output tokens. This game-changing insight allows them to scale their operations profitably without sacrificing model quality.

Staying Current with Weekly Model Releases

An AI researcher needs to stay updated on the latest model releases and benchmark scores for their work. By checking the "Pulse" feed on LLM Reference every week, they instantly see 177 new models, 53 price cuts, and 368 benchmark refreshes. This transformative feature eliminates the need to manually track dozens of sources, ensuring they always have the freshest data for their research and publications.

Routing Requests for Multi-Model Workflows

A platform engineer is building a system that routes different types of requests to different models. They use the "Don't want to pick? Route per request" feature to compare 17 verified gateways and routers. This allows them to design a transformative architecture that automatically selects the best provider, model, or tier for each incoming request, optimizing for cost, speed, and quality simultaneously.

How is the data on LLM Reference updated?

The data on LLM Reference is updated daily, with a major refresh cycle every week. This ensures that new model releases, verified price changes from providers, and benchmark score updates are captured in real-time. The "Pulse" feed highlights exactly what changed each week, including the number of new models, price cuts, and benchmark refreshes, so you are always working with the most current and game-changing information.

What are Editors' Picks and how are they determined?

Editors' Picks are curated recommendations for specific tasks like coding, agents, writing, and research. They are determined by analyzing the latest benchmark scores, real-world performance data, and community feedback. Each pick includes a detailed rationale, such as a specific benchmark score (e.g., 80.3% SWE-bench Pro), to explain why it is considered the best option. This transformative feature saves you time by providing a validated starting point for your model selection.

Can I compare two models directly on the platform?

Yes, the "Compare" feature allows you to select any two models and view a side-by-side comparison of their key attributes, including benchmark scores, pricing, context length, and provider details. This is a game-changing tool for making informed decisions, allowing you to directly pit leading models like "Claude Fable 5" against "GPT-5.5" to see which one best meets your specific needs and budget.

Is LLM Reference free to use?

Yes, LLM Reference is a free resource provided by the Data Advantage project. There are no subscription fees or paywalls for accessing the model directory, Editors' Picks, Pulse feed, or comparison tools. The platform is designed to be an essential, accessible resource for the entire AI community, unlocking the potential for everyone to make smarter, faster decisions about large language models.