Large Language Models

GPT-5.6

the flagship since 9 July 2026

Claude Fable 5

the flagship topping most July 2026 rankings

Claude Opus 5

frontier work with a dial on the bill

Claude Mythos 5

restricted twin of Fable 5

Claude Sonnet 5

the speed and intelligence balance

Gemini 3.5 family

the generation behind Gemini today

Muse Spark

Meta's proprietary frontier family

Gemini 3.6 Flash

the efficient workhorse

Grok 4.5

current xAI flagship, coding pivot

Kimi K3

the week's big story, largest open-weight model ever

Qwen3.8-Max

the first Qwen Max released with open weights

DeepSeek V4

the open-weight price disruptor

GLM-5.2

strongest open-weight all-rounder before K3 landed

Qwen 3.6

the multilingual open workhorse

Llama 4

the community's default base

Qwen3.7-Max

Qwen's premium, API-only tier

MiniMax M3

open-weight agentic climber

Mistral Large 3

Europe's flagship

Claude Opus 4.8

the benchmark everyone shoots at

Claude Sonnet 4.6

the superseded mid-tier workhorse

Claude Haiku 4.5

the fast cheap tier

GPT-5.5 "Spud"

recent predecessor, still widely deployed

OpenAI 2025-26 lineage (single roundup page, brief)

Kimi K2.6 and K2.7 Code

the run-up to K3

Nemotron 3

the transparency play

Gemma 4

the laptop model

Inkling

the newest lab's first big swing

Phi-4

small and on-device

Command R+ / Command A

retrieval and RAG for enterprises

Doubao / Seed

China's biggest consumer assistant

Ernie

the incumbent's answer

Hunyuan

the ecosystem model

Amazon Nova

the Bedrock native

Granite 4

compliance-first open models

Jamba

the SSM-transformer hybrid

Falcon

the Gulf's open flagship

OLMo

fully open science

LFM

the non-transformer bet

Sonar

search-native answering

Apple foundation models

the quiet on-device giant