Compare 55 models from 17 providers on price, context and licensing. Run the numbers with free calculators. Learn how it all works in plain English.
Prices and specs checked against provider documentation on October 10, 2026. No provider pays for placement.
Interactive demo: how a language model picks its next word
I need a language model for
This is how an LLM writes: it scores every possible next token, then samples one. Raise the temperature and the odds flatten; drop it to zero and the top token always wins. Pick a token to jump to the right page.
The newest models in the directory, each with specs and prices checked against the provider's own pages.
List prices per million tokens, and what a 30,000-conversation support bot would cost each month on each model.
| Model | Weights | Context | Input / 1M | Output / 1M | Support bot, per month* |
|---|---|---|---|---|---|
| GPT-6.1 SolOpenAI | Closed | 1.05M | $2 | $10 | $240 |
| Claude Sonnet 5.5Anthropic | Closed | 1M | $2 | $10 | $240 |
| Claude Opus 5.5Anthropic | Closed | 1M | $4 | $20 | $480 |
| Grok 4.7xAI | Closed | 500K | $2 | $6 | $192 |
| DeepSeek-V4.1-FlashDeepSeek | Open weights | 1M | $0.30 | $1.20 | $32 |
| GPT-6 AstraOpenAI | Closed | 1.05M | $10 | $50 | $1,200 |
| Gemini 3.8 FlashGoogle | Closed | 1M | $0.75 | $3.75 | $90 |
| Claude Fable 5.1Anthropic | Closed | 1M | $10 | $50 | $1,200 |
| GLM-5.3Z.ai (Zhipu AI) | Open weights | 1M | $1.40 | $4.40 | $137 |
| Qwen3.8-2.4T-A95BAlibaba (Qwen) | Open weights | 256K | $2 | $6 | $192 |
*30,000 conversations of 2,000 input and 400 output tokens at list price. Frontier and advanced models, newest first.
Seven calculators and builders. No sign-up and nothing to install; what you type stays in your browser.
Start from what you're building and get the models that fit, with prices and licenses side by side.
Our value index weighs capability, price, context, openness and recency. Move the sliders to match your priorities.
Head-to-head specs and the monthly bill for three real workloads.
Tell us what you're building. Within 2 business days you get a model shortlist, an architecture outline and a cost estimate, plus optional introductions to vetted implementation partners.
14 plain-English guides on how LLMs work, what they cost and where they fail, each checked against primary sources.
Hand-picked explainers from people who build these systems, each with a short summary so you know what you're getting.
A one-hour tour of what a large language model is, how one gets trained, and where it still falls short, presented without heavy math. It suits newcomers who want a complete mental model before reading about any single technique.
A short, animated primer on how a language model predicts the next word, how training tunes billions of parameters, and what makes transformers different. It works as a first stop for beginners and as a quick refresher for everyone else.
A multi-hour walkthrough of the full pipeline behind assistants like ChatGPT, from gathering web text and tokenizing it to pretraining, supervised fine-tuning and reinforcement learning. It rewards technical readers and curious builders who want a durable picture of how modern models are made and why they behave as they do.
The most recent milestones from our timeline of 65 events, from the 2017 Transformer paper to this month. Every entry links to a primary source.
OpenAI began rolling GPT-6 out across ChatGPT, including the free tier, and added Intelligent UI, which can answer with charts, forms and small interactive tools as well as text.
OpenAI opened its GPT-6 generation with Astra, which it said met its 'Critical' cybersecurity threshold, so it shipped with safeguards that refuse advanced cyber tasks such as proof-of-concept exploits.
Alibaba's Qwen3.8 open-weight releases included a 27B Apache 2.0 model and a 2.4-trillion-parameter mixture-of-experts flagship (95B active) under a separate custom license.
The Digital Omnibus amendments pushed high-risk AI obligations to December 2027 and August 2028, days before the Act's general application date of August 2, 2026.
Anthropic made a Mythos-class model generally available as Fable 5, routing some cyber, bio and distillation queries to Opus 4.8, while the less-restricted Mythos 5 went to vetted partners.
Unveiled at Google I/O, Gemini 3.5 Flash opened Google's 3.5 generation and was generally available in the Gemini API on day one; a larger 3.5 Pro was promised later.
LLModel is independent and reader-supported. Here's how to take part.
LLModel.com is an independent reference for large language models. It tracks the models from major providers and open-weight labs with prices, context windows and licenses checked against provider documentation, and adds free calculators, comparisons, plain-English guides, a glossary and curated videos.
Every model entry lists the official pages we checked and the date we checked them. We re-verify prices and specs against provider documentation on a regular schedule and when providers announce changes. Prices still change often, so confirm on the provider's pricing page before you buy.
It depends on the task, your budget, how much text you send, and whether data can leave your network. Start with the five-question model picker or a shortlist by job, then test the top two or three models on a sample of your own work and compare cost per correct answer.
Yes. The calculators, comparison tool, token counter and prompt builder are free and need no account. They run in your browser, so the text and numbers you enter aren't sent to us.
Through display advertising, clearly labeled sponsorships, optional introductions to implementation partners when you ask for help, and reader support. No provider pays to change a ranking, a score or an editorial verdict. Our editorial policy explains the rules.