A family is not one model
Compare a named release, not just a brand. Size, speed, modalities, and access can differ inside the same family.
THE AI LANDSCAPE 02 / MODELS & CAPABILITIES
Explore selected current AI releases: who makes them, what they focus on, and how you can access them. A model map with the context behind each name.
THE BIG PICTURE
Selected coverage. Check each entry for context and evidence.
Try a different name or a broader category.
A manually reviewed snapshot as of September 23, 2026. Colors organize the evidence; they do not score quality. On smaller screens, swipe across the comparison.
CONTEXT MAKES THE DIFFERENCE
Our interpretation of the comparison.
Compare a named release, not just a brand. Size, speed, modalities, and access can differ inside the same family.
An API release does not automatically appear in every consumer app. The access column follows the specific release announcement.
This map describes positioning and distribution. Vendor benchmark claims are not converted into a cross-provider performance ranking.
TRANSPARENT BY DESIGN
Official announcements, product documentation, and model repositories. All linked sources were checked on September 23, 2026. Publication dates and review dates are shown separately.
Selected general-purpose and specialist releases from major developers. Release context identifies an announcement, a dated changelog entry, or a current catalog; it is not always the original model launch date. Focus labels summarize the cited developer descriptions. Hosted means the cited release is accessed as a service; open weights means the specified checkpoint can be downloaded, subject to its license. Coverage is selective, not a complete release feed.
Dates on continuously updated documentation are labeled “Current documentation” unless a specific publication date is available. This page is maintained manually; the review date is not a live-feed timestamp. Company announcements describe the company’s claims and plans, not an independent audit.
OpenAI / Hosted models
Sol launched for the API, ChatGPT Work, and Codex on September 22. OpenAI said Chat access would follow. Astra remains a separate GPT-6 option.
OpenAI / Hosted models
Luna is the lighter GPT-6 variant. Its launch includes Free and Go desktop access as well as paid plans; access differs by product surface.
OpenAI / Hosted models
Astra is included alongside the newer Sol and Luna releases to distinguish a model family from its individual capability and efficiency tiers.
Anthropic / Hosted models
Anthropic announced Opus 5.5 across its supported platforms, including AWS, Google Cloud, and Azure. The Claude Platform identifier is claude-opus-5-5.
SpaceXAI / Hosted models
The September 21 announcement lists Cursor, Grok Build, and API distribution. This row does not assume simultaneous availability in every Grok app or plan.
Google DeepMind / Hosted models
Gemini 3.8 Flash is a general multimodal model. Flash Cyber is a separate specialized variant; its launch does not imply the same access conditions.
Google DeepMind / Hosted models
The API changelog records general availability of Gemini 3.8 Live and Live Extended Thinking on September 15. Live is a distinct interaction model, not simply a renamed Flash endpoint.
Meta / Hosted models
Meta lists Muse Spark 1.3 in its current developer catalog for coding and agentic work. The catalog check date is shown because it is not a dated launch announcement.
DeepSeek / Open weights
V4.1 Flash adds native visual understanding. DeepSeek publishes the checkpoint under MIT and serves it through the deepseek-flash alias. Downloadable weights still require substantial compute to run.
Qwen / Alibaba / Open weights
The official 27B checkpoint carries Apache 2.0. Other Qwen3.8 checkpoints can use different terms, so the license must be checked at the model level.