MODEL FAMILIES
Model family routes
Single-model pages capture narrow demand. Family pages capture the full route. Cluster the representative models together so people can keep moving deeper.
FAMILY PAGES
12
Break major model routes into standalone strong pages that can capture family-level search demand.
PROVIDERS
11
Family pages naturally connect provider paths, pricing layers, and the newest launches.
LATEST LINKS
9
Fresh models no longer sit alone. They now sit back inside the family route where decisions are easier.
DATASET NOTE
Model snapshot updated on 2026-05-11. Model-family pages cluster representative models, freshness signals, and provider paths into an entry that is easier to search and judge.
Family pages should make the newest branch changes obvious
Family pages are where new branches, previews, and generation shifts should be easiest to spot.
Latest radar
Claude Sonnet 4.6
Anthropic · Claude release signal
Global model routeClaudeInput price $3Context window 1M
Why it is worth opening now：Claude Sonnet 4.6 captures freshness-driven searches that already carry obvious intent, not…
See the new branch
Latest radar
Gemini 3.1 Pro Preview
Google · Gemini release signal
Global model routeGeminiInput price $2Context window 1.0M
Why it is worth opening now：Gemini 3.1 Pro Preview captures freshness-driven searches that already carry obvious intent…
See the new branch
Latest radar
DeepSeek R1 0528
DeepSeek · R1 release signal
China model routeR1Input price $0.45Context window 164K
Why it is worth opening now：DeepSeek R1 0528 captures freshness-driven searches that already carry obvious intent, not…
See the new branch
Latest radar
Grok 4.20
xAI · Grok release signal
Global model routeGrokInput price $3Context window 2M
Why it is worth opening now：Grok 4.20 captures freshness-driven searches that already carry obvious intent, not just br…
See the new branch
Start with these mainstream model-family routes
If you are choosing along a whole model line, these mainstream family routes get you moving faster than hopping model by model.
Start with onboarding
OpenAI
Start with the most common onboarding and commercial-readiness probe around OpenAI.
Open pilot page
Model-first route
Claude / Anthropic
Sample callable-model and failure-path signals around the Claude route first.
Open pilot page
Google route
Gemini / Google
Pull Gemini onboarding and route-readiness into the same early sampling lane.
Open pilot page
DeepSeek route
DeepSeek
Feed the high-frequency China-facing DeepSeek route with callable-model samples first.
Open pilot page
OpenAI GPT models
OpenAI’s core general-purpose route for copilots, complex tasks, and higher-value workflows.
GPT family1 models
Representative models
OpenAI o-series reasoning models
Built for complex reasoning, tool use, and agent workflows, with intent close to real technical decisions.
OpenAI reasoning family2 models
Claude models
Claude represents strong writing, knowledge workflows, and long-form analysis routes.
Claude family3 modelsLatest release
Representative models
Gemini models
Gemini naturally captures long-context, multimodal, and Google-ecosystem search intent.
Gemini family3 modelsLatest release
Representative models
Grok models
Grok carries both brand heat and freshness intent, which makes it ideal for routing buzz into deeper decisions.
Grok family3 modelsLatest release
Representative models
Mistral models
This family captures both general Mistral demand and narrower code or vision model searches.
Mistral model family4 modelsLatest release
Representative models
Llama models
Llama remains a core open-model route for directories, alternatives, and hosted-platform decisions.
Llama family3 models
Representative models
DeepSeek models
DeepSeek bridges China and global search interest, especially around pricing and reasoning.
DeepSeek family3 modelsLatest release
Representative models
Qwen models
The Qwen lineup is broad enough to connect general, low-cost, and multimodal intent on one page.
Qwen family4 modelsLatest release
Representative models
Kimi models
Kimi carries a broader mainstream audience in Chinese work, learning, and knowledge tasks.
Kimi family2 modelsLatest release
GLM models
GLM is a strong route for China enterprise procurement, reasoning, and multimodal decisions.
GLM family3 modelsLatest release
Representative models
MiniMax models
MiniMax fits long-context and enterprise-workflow searches, especially on China-side supplier routes.
MiniMax family2 modelsLatest release
Representative models