NO MODELS MATCH THE FILTER.
Models with a verified release date
Sorted by release date, newest first. Click a row to see the stories AINOW holds that mention the model.
-
Anthropic's latest Opus, with thinking on by default.
-
Alibaba's largest model, which challenged US flagships on benchmarks.
-
Moonshot's third generation with open weights and a million-token context window.
-
OpenAI's 5.6 series in three sizes: Sol, Terra and Luna.
-
xAI's frontier model with an emphasis on cost efficiency.
-
The Sonnet line's fifth generation, near-Opus quality on coding and agentic work.
-
Anthropic's most capable widely released model, with always-on thinking.
-
Announced alongside Fable 5. Restricted to Glasswing partners, not generally available.
-
DeepSeek's fourth generation with open weights and faster inference.
-
OpenAI's spring 2026 flagship, aimed at computer-based work.
-
The first Claude with high-resolution vision and task budgets.
-
OpenAI's most token-efficient reasoning model, with a 1M context window.
-
An upgrade to Gemini 3 with improved tool use.
-
Alibaba's Qwen3.5 generation with multimodal capability and open weights.
-
The first Claude with adaptive thinking and a 1M context window.
-
Introduced the effort parameter and substantially improved agentic work.
-
Google's third Gemini generation, with a large jump in reasoning.
-
OpenAI's fifth generation, with automatic routing between fast and deep modes.
-
Moonshot's open agentic model that competed with closed flagships.
-
xAI's fourth generation, with native tool use and real-time search.
-
The first Claude 4 generation, with a major lift on long-horizon coding.
-
The Qwen3 generation's flagship, with hybrid thinking modes.
-
Announced as the largest model in the Llama 4 family, but never made generally available.
-
An open reasoning model that ignited the debate on training cost and openness.
-
A Chinese MoE model trained for a fraction of Western budgets.
-
Google's agent-oriented generation with native tool use.
-
Alibaba's Qwen2.5 generation, which tightened China's grip on open weights.
-
The largest open model at launch, with 405 billion parameters.
-
The mid-tier model that beat the previous flagship — and introduced Artifacts.
-
OpenAI's omni model with speech, vision and text in one, at low latency.
-
Meta's third Llama generation, with markedly better instruction following.
-
The top model of the Claude 3 family, launched alongside Sonnet and Haiku.
-
Google's first Gemini generation, built multimodal from the ground up.
-
A small open model that beat far larger ones and accelerated European open source.
-
Meta's open-weights breakthrough: free for commercial use, with licence restrictions.
-
Anthropic's second-generation Claude, with a 100K context window at launch.
-
OpenAI's first multimodal flagship. Set the reasoning standard in 2023.
Without a verified release date
Models we have no checkable release date for. We never guess a date — they stay here until a source exists.
No model in the register is missing a release date right now.