MiniMax H3 Image released
MiniMax H3 Image is now in Image Gen with up to 4K output and 15 aspect ratios, and in Image Studio as MiniMax H3 Image Edit with up to 16 reference images.
New AI models, API improvements, and features, updated regularly.
MiniMax H3 Image is now in Image Gen with up to 4K output and 15 aspect ratios, and in Image Studio as MiniMax H3 Image Edit with up to 16 reference images.
ByteDance's Seedream v5.5, the newest Seedream model, is now in Image Gen with sharper photoreal detail, up to 2K output, 15 aspect ratios, PNG or JPEG files and native transparent backgrounds.
An uncensored build of Kimi K3.1 Pro joins the lineup: Moonshot AI's flagship reasoning model, with vision and a 1M-token context window.
An uncensored build of DeepSeek V4.1 Flash joins the lineup: DeepSeek's newest fast model, with quick replies for everyday chat, a 1M-token context window and provider zero data retention.
An uncensored build of Alibaba's Qwen 3.6 Plus joins the lineup, with a 1M-token context window for long conversations.
DeepSeek's newest fast model joins the lineup: the October 17 snapshot of V4.1 Flash, stronger at coding and agentic tasks with shorter answers, and a 1M-token context window.
GPT-5.6 Sol joins the lineup: the same underlying model as GPT-5.6 Sol Pro, served without the pro reasoning mode for faster everyday answers, with a 1M-token context window.
ByteDance's Seedream v5 Flash, the fast tier of the Seedream v5 family, is now in Image Gen, with up to 2K output, 15 aspect ratios and PNG or JPEG files.
xAI's newest flagship joins the lineup, succeeding Grok 4.6 with a larger base model that's better at long, multi-step tasks.
Anthropic's newest flagship joins the lineup, succeeding Claude Opus 5 with stronger reasoning, coding and long-form writing, plus a Roleplay version tuned for storytelling.
Anthropic's most capable model joins the lineup, built for deep reasoning, software engineering and long-form writing, with extended thinking and a 1M-token context window.
DeepSeek's newest flagship and its fast, low-cost counterpart are now in chat. V4 Pro is an open-weight reasoning model strong on maths and code; V4 Flash trades deliberation for speed. Both carry a 1M-token context window.
Alibaba's fast, low-cost open-weight model is now available in chat, tuned for quick replies and everyday conversation, with a 1M-token context window.
OpenAI's newest image models, GPT Image 2.5 Flare and GPT Image 2.5 Sunburst, are now in Image Gen. Flare is fast and sharp, Sunburst takes more time for finer detail, with up to 4K output, 15 aspect ratios, five quality levels and PNG, JPEG or WebP files.
OpenAI's most capable model joins the lineup, built for long-horizon reasoning, complex coding and deep research, with a 1M-token context window. Available on Premium and pay-as-you-go.
Google's newest high-speed thinking model, with significant gains over Gemini 3.7 Flash on software engineering, agentic tasks and multi-step reasoning, and a 1M-token context window.
GLM 5.3, Z.AI's newest open-weight reasoning model, is now available in chat. It succeeds GLM 5.2 with a 1M-token context window and stronger writing, roleplay and multi-step reasoning.
Anthropic's most advanced offering, improving on Claude Fable 5 across the board with the biggest gains in long-running agentic work and knowledge tasks.
xAI's newest flagship, stronger at reasoning and long conversations than Grok 4.5, which stays available alongside it.
A community-requested Claude Fable 5 variant tuned for dungeon-mastering and fantasy RPG play, with its own prompt and sampling.
Google's high-speed thinking model for fast agentic workflows, coding and multi-step reasoning, with a 1M-token context window.
A faster, cheaper Gemini Flash tier that ships alongside Gemini 3 Flash rather than replacing it.
Ten new image generation models join the lineup: Recraft V4 & V4 Pro, Seedream V4 & V5 Lite, GPT Image 1.5, Nano Banana 2, Flux 2 Max & Flex, Qwen Image 2 and ImagineArt 1.5 Pro. All at discounted pricing while it lasts.
Five more image models join the lineup at discounted deal pricing while it lasts.
Give any model its own avatar from Personalization. Models you skip keep using your default icon.
Open a side conversation on any message. Optionally update the original message from a thread reply.
A community-requested Kimi K3 variant tuned for roleplay and long-form writing, alongside the standard K3.
Moonshot AI's newest frontier model, a 2.8-trillion-parameter open-weight multimodal reasoning model with a 1-million-token context, built for complex coding and long-horizon agentic work.
xAI's newest flagship on its 1.5-trillion-parameter V9 foundation: Opus-class capability that's faster and more token-efficient.
OpenAI's highest-capability GPT-5.6 tier: pro-level reasoning with a 1M-token context window, built for hard, long-running tasks.
Anthropic's most advanced model, now uncensored, plus a community-requested Roleplay variant tuned for roleplay and writing.
A new uncensored Z-Image model, now available in image generation.
Organize your chats into folders to keep projects tidy.
ZAI's new roleplay model is now available in chat.
All Premium users can claim $5 in developer API credits from the API dashboard.
Free users now get more messages with free-tier chat models.
An experimental Claude Opus 4.8 variant tuned for roleplay and writing.
Manually add or auto-save memory items that carry across all your chats.
Create API keys, top up balance, and manage usage from the API dashboard.
Call models directly with custom parameters and system prompts.
xAI's reasoning model, available alongside Grok 4.20 Beta and Grok 4.
Three new models available in image generation.
Beta access to the Uncensored AI developer API.
Sync and restore your chats across devices.
ZAI's newest model, available in chat.
Community-requested variants tuned for roleplay and writing.
A short note about yourself, sent with every message to personalize responses across chats.
Long conversations now keep more memory and handle truncation more gracefully.
Claude Opus 4.7 started rolling out in chat.