FAQ
Answers are generated from the ledger checked on 31 Oct 2026.
Questions
How are prices sourced?
Prices, context limits, and release dates come only from official provider pricing pages, model cards, changelogs, or announcements. ModelMeter does not copy another aggregator. Each row stores the source link and the date that source was checked. If a value is not on the official page, it is left empty and shown as n/a. Each model row links to its source URL and shows the date that row was updated. The check date for the current ledger is 31 Oct 2026. The methodology is also written at https://modelmeter.live/about.
What does n/a mean?
Each row stores the source link and the date that source was checked. If a value is not on the official page, it is left empty and shown as n/a. ModelMeter does not estimate a missing price. A missing token price can be a rate the provider does not list on a public page, such as contact sales, or a price the provider removed when it deprecated the model. Rows with no published input rate or no published output rate: Command A+ (Cohere, status Active, price label "Not published"), Llama 3.3 70B (Groq, status Active, price label "Contact sales"), Mistral Large 3 (Mistral, status Deprecated, price label "Not published"), and Mistral Medium 3.5 (Mistral, status Deprecated, price label "Not published").
What do labels like 'limited', 'preview', 'deprecated', 'off-peak', and 'priced for prompts up to 100K' mean?
One number rarely describes every rate a provider charges. The figure in the table is the specific rate named on that row — standard, introductory, off-peak, or legacy. The note records the other published rates, such as long context, batch, peak hours, or a scheduled price change. Status and the price label are different fields. Limited access means the provider has not made the model generally available. The stored rate is still the published figure named in the price label. Current rows: Gemini 4 Carbon (Google) and GPT-6 Rosalind (OpenAI). Preview means the provider marks the model as a preview. Current rows: Mistral Large 4 (Mistral) and Muse Wonder (Meta). Deprecated means the provider has deprecated the model. The note records the retirement date and the replacement when the official page states them. Current rows: Mistral Large 3 (Mistral) and Mistral Medium 3.5 (Mistral). An off-peak price label means the stored input and output rates are the off-peak figures. The note records the peak figures when the official page published them. Current price label: "Off-peak" on DeepSeek Flash (DeepSeek) and DeepSeek V4 Pro (DeepSeek). A price label that names a prompt-length band means the stored rate applies to that band. The note records the other band when the official page published it. Current price label: "Prompts up to 100K tokens" on Claude Haiku 5.5 (Anthropic).
Why is a hosted open model listed under Together, Groq, or Mistral?
Hosted open models are labelled with the company that bills the API. A Together or Groq price is not a Meta list price. weightsFrom shows the lab that trained the weights. Current hosted rows: Llama 3.3 70B is billed by Groq and the weights are from Meta, GLM 5.3 is billed by Mistral and the weights are from Z.ai, GLM 5.3 Flash is billed by Mistral and the weights are from Z.ai, Muse Glimmer 30B is billed by Together AI and the weights are from Meta, and Muse Spark Open is billed by Together AI and the weights are from Meta.
Which is the cheapest model right now?
The lowest stored input rate is $0.10 per 1M tokens, on Claude Haiku 5.5 (Anthropic, Prompts up to 100K tokens, $0.10) and GPT-6 Luna (OpenAI, Standard, under 272K, $0.10). The lowest stored output rate is $0.50 per 1M tokens, on Claude Haiku 5.5 (Anthropic, Prompts up to 100K tokens, $0.50), GLM 5.3 Flash (Mistral, Standard, $0.50), and GPT-6 Luna (OpenAI, Standard, under 272K, $0.50). Rows with no published rate for that column are left out of the ranking. n/a is not treated as zero. The full rankings are at https://modelmeter.live/cheapest-ai-api.
Which model has the largest context window?
The largest published context window is 1.05M tokens, on GPT-6 Astra (OpenAI), GPT-6 Luna (OpenAI), GPT-6.1 Sol (OpenAI), and GPT-6.1 Terra (OpenAI). 2 models have no published context window and are left out. The full ranking is at https://modelmeter.live/largest-context-window-ai-models.
How often is data checked?
The dataset records the day it was checked. The current check date is 31 Oct 2026. Every row's updated date is 31 Oct 2026. That date is the day the official pages were re-read. It is shown on the home page, the about page, and each model row.
How do I follow price changes?
Dated price changes, new models, and scheduled rates are at https://modelmeter.live/price-changes. The RSS feed is https://modelmeter.live/price-changes.xml. Machine-readable rows for the whole ledger are at https://modelmeter.live/api/models.json.
Are any listings sponsored?
Sponsored placements are always labelled. When a sponsored placement is shown, it carries a visible Sponsored label. Sponsored slots are switched off, so this version shows no sponsored listing.