NeuralStackA new tool is trying to bring order to the rapid-fire pace of large language model launches. AI Release Tracker, a free site focused on frontier AI systems, maintains a continuously updated timeline of every major AI model release from leading labs, starting with the launch of ChatGPT on November 30 2022. As of mid-2026, it tracks 229 models from 11 companies and presents them in a single view that spans multiple years of development.
The timeline stretches from late 2022 through 2026, capturing month-by-month releases from OpenAI, Anthropic, Google, Meta, SpaceXAI, DeepSeek, Mistral, Moonshot AI, Cursor, Z.ai, and Qwen. Each company’s models are listed in sequence, with OpenAI’s GPT-3.5 and GPT-4 series giving way to the GPT-5 family, Anthropic’s Claude line progressing through multiple Sonnet, Haiku, Opus, and Fable iterations, and Google’s Bard and Gemini models evolving across several generations. Similar multi-step progressions are visible for Meta’s LLaMA family, SpaceXAI’s Grok models, DeepSeek’s V- and R-series, and newer players like Qwen, which has advanced its Qwen3 line through several releases.
The site does more than simply list names. Every individual model page includes core technical details such as release date, parameter count, context window, model type, and benchmark scores where available, spanning tests like GPQA Diamond for graduate-level science reasoning, SWE-Bench Verified for real-world software engineering, and MMMU for multimodal tasks. An analytics section aggregates those datapoints into cumulative release charts, monthly release frequency, year-over-year totals, benchmark frontier progression, and release cadence broken down by company. In effect, the tracker turns a chaotic release landscape into something more measurable.
AI Release Tracker’s data has already been picked up outside the AI community. Business Insider has cited the site’s analysis showing that the monthly cadence of major AI releases has roughly quadrupled since 2023, underscoring just how quickly the frontier is shifting. That acceleration is visible in the dense clusters of model names across the timeline, and in the way each lab now maintains distinct families and sub-families of models for different use cases, from coding assistants to high-context reasoning systems.
At the benchmark level, the tracker calls out which systems currently sit at the top of key evaluations. On the GPQA Diamond test of graduate-level science reasoning, the current leader is OpenAI’s GPT-5.4-Pro. On SWE-Bench Verified, which focuses on real-world software engineering tasks, Anthropic’s Claude Opus 4.7 holds the top spot. These leaders are flagged on the analytics page, allowing observers to see not only who is shipping models fastest, but which releases are defining the cutting edge in specific domains.
Even as it focuses on long-term trends, the site is also tuned to near-term developments. For each of the tracked companies, AI Release Tracker lists expected upcoming releases, including dates for OpenAI, Anthropic, Google, Meta, SpaceXAI, DeepSeek, Mistral, Moonshot AI, Cursor, Z.ai, and Qwen later in 2026. The homepage highlights the most recent frontier models, and the FAQ notes that the latest tracked release is GLM-5.3 by Z.ai, updated on August 14 2026. That combination of historical context and forward-looking scheduling turns the tracker into a reference both for what has happened and what is about to arrive.
Why this matters
The pace and opacity of AI model releases have made it difficult for policymakers, researchers, and users to understand who is deploying what, and how quickly capabilities are advancing. By cataloging every major frontier model across 11 labs, and pairing timelines with live benchmark leaders and release cadence analytics, AI Release Tracker offers a clearer empirical view of the industry’s trajectory. In a market where monthly launch frequency has reportedly quadrupled since 2023, that kind of structured visibility can inform debates on regulation, investment, and risk, and help distinguish sustained progress from hype.
Looking ahead, AI Release Tracker’s utility is likely to grow as the frontier becomes more crowded and specialized. As OpenAI, Anthropic, Google, Meta, SpaceXAI, DeepSeek, Mistral, Moonshot AI, Cursor, Z.ai, and Qwen continue to roll out new models, the site’s cumulative charts and benchmark tracking will provide an ongoing record of how different strategies play out over time. If the release tempo continues to accelerate, tools like this may become essential infrastructure for anyone trying to stay oriented in the AI ecosystem, from engineers and researchers to journalists and regulators watching the next wave of frontier systems approach.