AI model release timeline

GLM-5.3-Flash launched with MIT-licensed open weights.

Qwen3.8-Flash-Next released open weights and became available as managed Qwen3.8-Flash.

DeepSeek-V4-Flash-Vision-Exp launched in the multimodal API.

GLM-5.3 launched with gains attributed entirely to scaled post-training.

Qwen3.8-27B weights were released.

DeepSeek-V4-Pro-0813 reached general availability with adjustable reasoning effort.

Gemini 3.7 Flash reached general availability.

Grok 4.6 launched for long-running agents and complex visual work.

Qwen3.8-2.4T-A95B weights were released.

Meta released the open local-agent model Muse Glimmer 30B.

Wan 3.0 Video entered preview.

Muse Spark 1.2 and Muse Code launched.

MiniMax opened H3 weights.

Qwen3.8-Max launched.

DeepSeek-V4-Flash-0731 entered public beta in the API.

Seedance 2.5 launched.

MiniMax launched the H3 omni-modal video model.

Claude Opus 5 launched.

Midjourney V8.2 became the default model.

Gemini 3.5 Flash-Lite reached general availability.

Gemini 3.6 Flash reached general availability.

Qwen-Image 3.0 launched.

Kimi K3 was announced at 2.8T parameters with native vision.

Grok 4.5 launched in Grok Build, Cursor, and the API.

GPT-5.6 reached general availability across ChatGPT, Codex, and API.

Muse Spark 1.1 entered public preview through Meta Model API.

Seedream 5.0 Pro launched.

Muse Image launched.

Meta previewed Muse Video.

Tencent formally launched and opened Hy3.

Claude Sonnet 5 launched and became the default for Free and Pro.

Nano Banana 2 Lite launched.

Gemini Omni Flash opened video-generation API access.

GPT-5.6 Sol, Terra, and Luna entered limited preview.

Xiaomi opened MiMo-V2.5 series weights.

Seed2.1 officially launched.

GLM-5.2 launched with open weights.

Google released experimental DiffusionGemma.

Claude Fable 5 reached general availability alongside Mythos 5.

Luma launched Ray3.2 and its API.

Gemma 4 12B launched.

MiniMax launched the native multimodal 1M-context M3.

Claude Opus 4.8 launched.

Google announced Gemini 3.5 Flash at I/O.

Gemini 3.1 Flash-Lite reached general availability.

GPT-5.5 Instant updated ChatGPT’s default fast experience.

The Sora product was discontinued.

DeepSeek released V4-Pro Preview with open weights.

GPT-5.5 and GPT-5.5 Pro became available in the API.

DeepSeek-V4-Flash Preview launched with V4-Pro Preview and open weights.

Tencent released Hy3 Preview.

GPT-5.5 launched in ChatGPT and Codex.

MiMo-V2.5 entered public beta.

MiMo-V2.5-Pro entered public beta.

Qwen3.6-27B released open weights.

GPT Image 2 and ChatGPT Images 2.0 launched.

Kimi K2.6 launched with open weights.

Claude Opus 4.7 reached general availability.

Qwen3.6-35B-A3B released open weights.

Midjourney V8.1 Alpha launched.

Meta introduced Muse Spark, the first Muse model.

GLM-5.1 launched.

Wan 2.7 audio and multi-reference video models launched.

Gemma 4 31B launched.

Gemma 4 26B MoE launched.

Wan 2.7 Image launched.

MiniMax-M2.7 launched.

Xiaomi launched the trillion-parameter MiMo-V2-Pro.

MiMo-V2-Omni launched.

GPT-5.4 mini launched for coding and subagents.

GPT-5.4 nano launched in the API.

GPT-5.4 Thinking and Pro launched in ChatGPT, API, and Codex.

GPT-5.3 Instant became ChatGPT’s everyday fast model and entered the API.

Nano Banana 2 entered the Gemini API.

Gemini 3.1 Pro Preview launched.

Claude Sonnet 4.6 launched.

Qwen3.5-397B-A17B opened as the first Qwen3.5 release.

The Seed2.0 series launched.

Seedream 5.0 Lite launched.

GLM-5 launched for complex systems engineering and long-horizon agents.

GPT-5.3-Codex-Spark launched as a real-time coding research preview.

MiniMax-M2.5 launched.

Seedance 2.0 launched.

Qwen-Image 2.0 launched.

GPT-5.3-Codex launched across Codex products.

Claude Opus 4.6 launched.

Kuaishou launched the Kling AI 3.0 multimodal line.

Kimi K2.5 launched with open weights.

Ray3.14 launched.

FLUX.2 klein 4B was released.

MiniMax-M2.1 launched.

GLM-4.7 launched with open weights.

GPT-5.2-Codex launched for long-horizon software engineering and defensive security.

Seed1.8 general agent model launched.

Gemini 3 Flash Preview launched.

MiMo-V2-Flash launched.

GPT Image 1.5 launched.

Seedance 1.5 Pro launched.

GPT-5.2 Instant, Thinking, and Pro launched.

DeepSeek-V3.2 launched and upgraded the chat and reasoner API aliases.

DeepSeek-V3.2-Speciale opened as the high-reasoning variant.

Runway Gen-4.5 launched.

Kling 2.6 and O1 launched.

The FLUX.2 family launched.

Claude Opus 4.5 launched with effort control.

Nano Banana Pro launched.

Grok 4.1 Fast and the Agent Tools API launched.

Gemini 3 Pro Preview launched.

Grok 4.1 launched for all Grok users.

GPT-5.1 launched in the API with adaptive reasoning and new developer tools.

Kimi K2 Thinking launched with open weights.

Hailuo 2.3 and its Fast variant launched.

MiniMax-M2 launched.

Claude Haiku 4.5 launched.

Veo 3.1 and new Flow editing tools launched.

OpenAI launched Sora 2.

Claude Sonnet 4.5 launched.

Tencent opened HunyuanImage 3.0.

Qwen3-Max launched.

Grok 4 Fast launched.

Luma launched the visual-reasoning video model Ray3.

Qwen3-Next-80B-A3B released open weights.

Seedream 4.0 launched.

Gemini 2.5 Flash Image launched as Nano Banana.

GPT-5 launched in ChatGPT and the API.

GPT-5 mini launched as the lower-cost API tier.

GPT-5 nano launched as the smallest API tier.

Qwen-Image 20B weights were released.

GLM-4.5 and GLM-4.5-Air launched with open weights.

Wan 2.2 released its open MoE video models.

Kimi K2 launched as an open 1T MoE agent model.

Grok 4 and Grok 4 Heavy launched.

Gemma 3n fully released its edge multimodal models.

Seed1.6 launched.

Midjourney launched its first image-to-video model.

MiniMax Hailuo 02 launched.

MiniMax-M1 opened its 1M-context reasoning weights.

Seedance 1.0 launched.

FLUX.1 Kontext launched.

Claude Opus 4 launched alongside Sonnet 4.

Claude Sonnet 4 launched as the balanced Claude 4 tier.

Google launched Imagen 4.

Veo 3 added native audio.

Xiaomi released MiMo-7B-RL.

Qwen3 launched with the 235B-A22B MoE as flagship.

GPT Image 1 entered the API.

Llama 4 Scout launched as an open multimodal MoE.

Llama 4 Maverick launched alongside Scout.

Midjourney V7 entered alpha.

Runway Gen-4 launched.

Gemini 2.5 Pro Experimental launched with thinking built into the new generation.

Gemma 3 added vision and 128K context.

Wan 2.1 weights and inference code were released.

Claude 3.7 Sonnet launched as a hybrid-reasoning model.

Grok 3 Beta was unveiled as a reasoning-agent generation.

Gemini 2.0 Flash became generally available through the API and Vertex AI.

DeepSeek-R1 and its distilled models were released with open weights.

Moonshot published its Kimi K1.5 multimodal reasoning work.

Luma launched Ray2.

DeepSeek-V3 launched as an open 671B MoE.

Google launched Veo 2.

Sora Turbo launched to users.

Llama 3.3 70B launched.

Tencent opened Hunyuan Large weights.

The Stable Diffusion 3.5 series was released.

Llama 3.2 added vision and edge models.

Imagen 3 became available to users.

Black Forest Labs launched FLUX.1.

Llama 3.1 405B and 128K context became available.

Gemma 2 27B was released.

Stable Diffusion 3 Medium was released.

The open Qwen2 family launched with a 72B flagship.

GPT-4o launched with native text, vision, and audio.

DeepSeek-V2 released model weights and its technical report.

Meta released Llama 3 8B and 70B.

The Claude 3 family launched with Opus as the flagship tier.

Google released the first Gemma open models.

Gemini 1.5 Pro entered limited preview with a 1M context window.

Midjourney V6 launched.

Google announced the natively multimodal Gemini 1.0 family.

Qwen-7B launched the Qwen open-model line.

SDXL 1.0 was released.

Claude 2 launched through the API and claude.ai beta.

GPT-4 launched as a large multimodal model.

Anthropic formally introduced Claude.

Stable Diffusion was publicly released with weights.

The GPT-3 paper documented few-shot learning at 175B parameters.

The full 1.5B GPT-2 weights were released.

OpenAI announced GPT-2 and began a staged release.