Catalogue
Models
455 models, and deliberately not every model there is: one earns a place by being sold by a provider we track, measured by a leaderboard we follow, or published as open weights by a lab we follow. Pointer names like “-latest” stay out, because they answer for whatever the vendor aimed them at that week.
How we choose what to list14 models
Muse Spark5 sizes here
Muse Spark 1.3 ContributorMetaMuse Spark 1.3 Contributor is a hosted-only model: we list no download for it, so using it means choosing a host.writingnot measuredcodingnot measuredagentsnot measuredeverydaynot measured$0.10 / $0.20per 1Mvia MetaMuse Spark 1.3MetaMuse Spark 1.3 is Meta's proprietary multimodal model that can handle up to one million tokens in a single request.writing26th of 168 coding7th of 168 agents15th of 55 everyday7th of 168 $1.25 / $4.25per 1Mvia MetaMuse Spark 1.2 ContributorMetaMuse Spark 1.2 Contributor is a hosted-only model: we list no download for it, so using it means choosing a host.writingnot measuredcodingnot measuredagentsnot measuredeverydaynot measured$0.10 / $0.20per 1Mvia MetaMuse Spark 1.2MetaMuse Spark 1.2 is a hosted-only model from Meta, so using it means choosing a host rather than running it yourself.writing28th of 168 coding8th of 168 agents37th of 55 everyday5th of 168 $1.25 / $4.25per 1Mvia MetaMuse Spark 1.1MetaMuse Spark 1.1 is Meta's hosted-only multimodal model that can handle up to one million tokens in a single request.writing14th of 168 coding10th of 168 agents39th of 55 everyday10th of 168 $1.25 / $4.25per 1Mvia MetaMuse Glimmer 30BMetaMuse Glimmer 30B is a 29.8-billion-parameter text-and-image model from Meta with a permissive Apache licence.writing92nd of 168 coding69th of 168 agentsnot measuredeveryday75th of 168 $0.30 / $1.10per 1Mvia PhalaLlama Guard 4 12BMetaLlama Guard 4 12B is a safety-classifier model from Meta that reads text and images to judge whether content is safe.writingnot measuredcodingnot measuredagentsnot measuredeverydaynot measured$0.18 / $0.18per 1Mvia OpenRouter
Llama7 sizes here
Llama 4 ScoutMetaLlama 4 Scout is a downloadable model with a very large request capacity and image input, but it sits near the bottom of the field on the dated boards we can quote, and its licence puts conditions on commercial use. Try it through a host before committing to a download.writing134th of 168 coding135th of 168 agentsnot measuredeveryday137th of 168 $0.10 / $0.30per 1Mvia OpenRouterLlama 4 MaverickMetaLlama 4 Maverick is a downloadable model you can run yourself, with text and image input and room for long documents in one request.writing124th of 168 coding131st of 168 agentsnot measuredeveryday134th of 168 $0.19 / $0.65per 1Mvia OpenRouterLlama 3.3 70B InstructMetaLlama 3.3 is a 70.6-billion-parameter text model from Meta with a 131,072-token request limit and a wide spread of hosted providers.writing136th of 168 coding146th of 168 agentsnot measuredeveryday138th of 168 $0.10 / $0.32per 1Mvia OpenRouterLlama 3.2 3B InstructMetaLlama 3.2 is a tiny downloadable model from Meta built for on-device and low-latency tasks.writing167th of 168 coding167th of 168 agentsnot measuredeveryday167th of 168 $0.050 / $0.33per 1Mvia OpenRouterLlama 3.2 1B InstructMetaLlama 3.2 is Meta's smallest instruction-tuned model, built for edge devices and cost-sensitive tasks where a tiny footprint matters more than top-tier capability.writing168th of 168 coding168th of 168 agentsnot measuredeveryday168th of 168 $0.020 / $0.020per 1Mvia Novita AILlama 3.1 8B InstructMetaLlama 3.1 8B Instruct is a small text model you can download and run on a single modern graphics card, or reach cheaply through a host.writing166th of 168 coding166th of 168 agentsnot measuredeveryday166th of 168 $0.050 / $0.080per 1Mvia GroqLlama 3.1 70B InstructMetaLlama 3.1 is a 70.6-billion-parameter text model from Meta released in 2024 with a 131,072-token request limit.writing148th of 168 coding149th of 168 agentsnot measuredeveryday149th of 168 $0.40 / $0.40per 1Mvia OpenRouter