AI · Reference · continually updated
Large Language Models
The models that read, reason and write, from the frontier labs to the open-weight field.
- models
- 54
- labs
- 29
- newest model
- Gemini 3.8 Live15 Sept 2026
- guides
- 6
- Gemini 3.8 Live · 15 Sept 2026
- DeepSeek-V4.1-Flash · 10 Sept 2026
- GPT-6 Astra · 3 Sept 2026
- K2 Horizon · 3 Sept 2026
- Muse Spark · 2 Sept 2026
- Gemini 3.8 Flash · 2 Sept 2026
- Claude Fable 5.1 · 1 Sept 2026
- Hy4 preview · 27 Aug 2026
- Qwen3.8-Flash-Next · 26 Aug 2026
- GLM-5.3-Flash · 26 Aug 2026
- DeepSeek-V4-Flash-Vision-Exp · 21 Aug 2026
- Ox Alpha · 20 Aug 2026
Newest first · 54 models · continually updated
Latest LLMs
just in: Gemini 3.8 Live · 15 Sept 2026
Most recent and top modelsAll 54 models · newest first
Google DeepMindGemini 3.8 Live
Released 15 Sept 2026the live dialogue pair, one built for scale and one for thinking
DeepSeekDeepSeek-V4.1-Flash
Released 10 Sept 2026a 552B open-weight model built around its cache
OpenAIGPT-6 Astra
Released 3 Sept 2026the new generation, rolling out from 3 September 2026
MBZUAI IFMK2 Horizon
Released 3 Sept 2026a six-model Apache 2.0 fleet released with its checkpoints, logs and datasets
MetaMuse Spark
Released 2 Sept 2026Meta's proprietary frontier family
Google DeepMindGemini 3.8 Flash
Released 2 Sept 2026the workhorse again, plus a cyber variant behind a gate
AnthropicClaude Fable 5.1
Released 1 Sept 2026built for the API, hard on a subscription allowance
TencentHy4 preview
Released 27 Aug 2026an open-weight flagship on a standard open-source licence
AlibabaQwen3.8-Flash-Next
Released 26 Aug 2026the cheap tier, and a preview of what Qwen4 is built on
Z.ai / ZhipuGLM-5.3-Flash
Released 26 Aug 2026a new base, natively multimodal, at a tenth of GLM-5.2's price
DeepSeekDeepSeek-V4-Flash-Vision-Exp
Released 21 Aug 2026an experimental vision model with no weights behind it
Anonymous (Stealth)Ox Alpha
Released 20 Aug 2026an anonymous listing, measured rather than guessed
Z.ai / ZhipuGLM-5.3
Released 14 Aug 2026a post-training update with an emergent cyber capability
Google DeepMindGemini 3.7 Flash
Released 13 Aug 2026the workhorse tier, on an introductory price
DeepSeekDeepSeek V4 Pro
Released 13 Aug 2026the price disruptor, raising its prices
xAIGrok 4.6
Released 12 Aug 2026xAI's current flagship, built for long-running agents
Liquid AILFM
Released 4 Aug 2026the non-transformer bet
AlibabaQwen3.8-Max
Released 3 Aug 2026the first Qwen Max released with open weights
AnthropicTop modelClaude Opus 5
Released 24 Jul 2026frontier work with a dial on the bill
Google DeepMindGemini 3.5 family
Released 21 Jul 2026the generation behind Gemini today
Google DeepMindGemini 3.6 Flash
Released 21 Jul 2026the efficient workhorse
Moonshot AIKimi K3
Released 17 Jul 2026the largest open-weight model yet released
Thinking Machines LabInkling
Released 15 Jul 2026the newest lab's first big swing
OpenAITop modelGPT-5.6
Released 9 Jul 2026the flagship from 9 July to 3 September 2026
xAIGrok 4.5
Released 8 Jul 2026the coding pivot, superseded by Grok 4.6
TencentHunyuan
Released 6 Jul 2026the ecosystem model
AnthropicTop modelClaude Sonnet 5
Released 30 Jun 2026the speed and intelligence balance
ByteDanceDoubao / Seed
Released 23 Jun 2026China's biggest consumer assistant
Z.ai / ZhipuGLM-5.2
Released 13 Jun 2026strongest open-weight all-rounder before K3 landed
MoonshotKimi K2.6 and K2.7 Code
Released 12 Jun 2026the run-up to K3
AnthropicTop modelClaude Fable 5
Released 9 Jun 2026the flagship holding first place on the Arena text board
AnthropicTop modelClaude Mythos 5
Released 9 Jun 2026restricted twin of Fable 5
Apple foundation models
Released 8 Jun 2026the quiet on-device giant
NVIDIANemotron 3
Released 4 Jun 2026the transparency play
MiniMaxMiniMax M3
Released 1 Jun 2026open-weight agentic climber
AnthropicClaude Opus 4.8
Released 28 May 2026the benchmark everyone shoots at
AlibabaQwen3.7-Max
Released 21 May 2026Qwen's premium, API-only tier
CohereCommand R+ / Command A
Released 20 May 2026retrieval and RAG for enterprises
BaiduErnie
Released 9 May 2026the incumbent's answer
TII, UAEFalcon
Released 3 May 2026the Gulf's open flagship
IBMGranite 4
Released 29 Apr 2026compliance-first open models
DeepSeekDeepSeek V4
Released 24 Apr 2026the open-weight price disruptor
OpenAIGPT-5.5 "Spud"
Released 23 Apr 2026recent predecessor, still widely deployed
AlibabaQwen 3.6
Released 22 Apr 2026the multilingual open workhorse
GoogleGemma 4
Released 2 Apr 2026the laptop model
MicrosoftPhi-4
Released 4 Mar 2026small and on-device
AnthropicClaude Sonnet 4.6
Released 17 Feb 2026the superseded mid-tier workhorse
Mistral AIMistral Large 3
Released 2 Dec 2025Europe's flagship
AWSAmazon Nova
Released 2 Dec 2025the Bedrock native
Ai2OLMo
Released 20 Nov 2025fully open science
AnthropicClaude Haiku 4.5
Released 15 Oct 2025the fast cheap tier
AI21Jamba
Released 8 Oct 2025the SSM-transformer hybrid
MetaLlama 4
Released 5 Apr 2025the community's default base
PerplexitySonar
Released 11 Feb 2025search-native answering
Guides and roundups
AI model fingerprints
counting tokens across 278 models to see which ones share a lineage
AnthropicPrompting Claude Fable 5
the controls Fable 5 gives you, and the ones it takes away
AnthropicPrompting Claude Opus 5
Anthropic's agentic-coding default, and how to keep it in bounds
Moonshot AIPrompting Kimi K3
the open-weight flagship, on Moonshot's API and on your own hardware
OpenAIPrompting GPT-5.6 Sol
OpenAI's frontier tier, and the settings that decide what it costs
OpenAI 2025-26 lineage