Header

• Last Updated : 1/4/2026
• Last Updater : lilhuss26

Intro

• Purpose : This file's goal Isn't collecting ALL models in one place, but quite the opposite, is kind of filtering. All mentioned models/providers will be latest OR actually used ones, that deserve mentioning. Making it a quick reference for any future projects.
• To Mention : ArtificialAnalysis, U can find more stats and comparison

TOC

• The Big Three
  • Claude
  • Gemini
  • OpenAI
• Families
  • Mistral3
  • Inception
  • Z.ai
  • Qwen
• Providers
  • OpenRouter
  • Groq

Legend

• Pricing in any table is 1M token input cost/1M token output cost
• Throughput is tps (Tkens Per Second)

The Big Three

Claude

Note : Claude provides some older versions, with the same pricing, Except Haiku , Significantly cheaper, but with very low performance on the benchmarks.

Name	Cat	Pricing	Tried?	ToolCalling	Throughput	Performance	Notes
Haiku 4.5	Meduim	1/5	No
Sonnet 4.6	Large	3/15	No
Opus 4.6	Super	5/25	No

OpenAI

Name	Cat	Pricing	Tried?	ToolCalling	Throughput	Performance	Notes
Chat-GPT 4o	Large		Yes	True		Great, large prompt (deep analysis)
Chat-GPT 5	Large		Yes			Good	Slower, and it's output feels too soft

---

Families

Mistral3

Name	Cat	Pricing	Tried?	ToolCalling	Throughput	Performance	Notes
Mistral3-largest	Meduim	0.5/1.5	Yes	True	30	Great, Agent stable	SLOW
Mistral3-24b	Small		No	True
Mistral3-14b	Small	0.2/0.2	Yes	True	98	Agent stable, large prompts (Re-writting)	Small hullicantaion

Inception

Name	Cat	Pricing	Tried?	ToolCalling	Throughput	Performance	Notes
Mercury2	Meduim	0.25/0.75	Yes	True	332	Great, Agent very smart and stable

---

Providers

OpenRouter

Note : It's wide models coverage and big trial qouta, makes it the first option for exploration

Models variaty	Free trial	Used	Notes
All Models	Enough	Yes	Free trial is random, stops at the same day OR Last for weeks

Groq

Models variety	Free trial	Yes	Notes
Small Collection	Enough	No