Know the model.
Test the workflow.

1. 01

   Confirm identity and access

   Verify the provider, exact model label, region, account context, input modes, and visible limitations before drawing a conclusion.

2. 02

   Use a repeatable shot set

   Test a small range of representative shots with fixed inputs and review criteria. Keep observation separate from preference.

3. 03

   Document the evaluation date

   Interfaces, model versions, and capabilities evolve. A dated note is more useful than an undated superlative.

4. 04

   Model query guide: choose a query by decision

   The English Model Hub indexes 433 detail queries as an editorial decision directory, not as a live inventory of models available through SEELE. Each card preserves the exact query and exposes its supporting review topics. Start with the wording closest to the question you need to answer, then use the reader intent and taxonomy job to decide whether the page is about identity, an exact version, access, a host or application, a capability, an output limit, deployment, policy, troubleshooting, comparison, or provenance. A card title can contain a provider-like name, number, platform, or entitlement without confirming any of them. Open the detail boundary before planning a test, purchase, download, integration, or production handoff.

5. 05

   Model query guide: directory map by reader intent

   50 entries address access and entitlement review; representative queries are “seedance 2.0 free unlimited”, “transcrever video em texto gratis”, and “veo 3 مجاني”. Use that group to determine whether the reader means discovery access, a trial, a recurring allowance, a paid plan, a downloadable package, an open license, or commercial-use permission. 16 entries address availability and symptom diagnosis; representative queries are “sora 2 pc requirements”, “higgsfield down”, and “sora 2 low quality”. Use that group to separate an outage report, account restriction, unsupported input, output limit, performance complaint, duration request, and quality observation. 31 entries address capability-scope review; representative queries are “ltx-video”, “ltx ai video”, and “grok video”. Use that group to translate the query into a bounded input, control, and observable output without assuming that a broad product label supports the requested operation. 38 entries address provider and distribution-surface review; representative queries are “ltx.studio”, “ltx studios”, and “gemini vidoe”. Use that group to identify separately the model developer, product operator, host, reseller, application publisher, and account surface named or implied by the request. 7 entries address model-identity resolution; representative queries are “beatrice v2”, “seedream”, and “google banana nano”. Use that group to resolve whether the phrase is an official model, product, feature, alias, misspelling, translation, unreleased label, or unrelated entity before describing behavior. 12 entries address integration and deployment review; representative queries are “comfyui sora 2”, “run wan 2.5 locally”, and “nano banana pro local”. Use that group to separate hosted API use, local weights, workflow nodes, protocol adapters, cloud images, and third-party integrations because their evidence and security boundaries differ. 133 entries address model orientation and fit review; representative queries are “sora 2 videos”, “leonardo нейросеть”, and “jimeng ai english”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 3 entries address bounded model evaluation; representative queries are “sora ai image”, “sora image generation”, and “kling motion”. Use that group to define the requested media task and acceptance criteria before deciding whether the named model or feature is the correct comparison unit. 17 entries address safety and policy review; representative queries are “sora refugees”, “wan 2.5 unrestricted”, and “wan 2.2 spicy”. Use that group to identify the legitimate editorial or diagnostic question without converting sensitive wording into evasion, non-consensual identity use, or claims about safeguards. 6 entries address model orientation and fit review; representative queries are “Runway Ruby HDR conversion model evaluation”, “LongCat-Avatar talking video model evaluation”, and “MiniMax Design motion graphics model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 AI Edit model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “CapCut Seedance 2.5 Multi-Track model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 editing model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 storyboard model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 typography model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 motion graphics model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 AI Extend model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance AI influencer vlog model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 1080p model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 2K video model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 local rendering model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Google Flow + King AI model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “AI generated worlds model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 one continuous shot model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 UGC ad model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 native 4K model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Max model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 native audio model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 Skill: Script to Video model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “agent-native video workflow model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Wan 3.0 model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax Design model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Flova x Seedance 2.5 model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 Script to Video skill model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 6 entries address model orientation and fit review; representative queries are “Wan 3.0 on Topview model identity check”, “Wan 3.0 on Newtake model identity check”, and “Wan 3.0 on Flova model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.0 Fast CapCut model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax Design comic-book page Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Max launch pricing Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Max Turbo Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 multi-shot prompt Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Fotor Video Agent Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Wan 3.0 video editing Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 Mandarin prompts Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “video benchmarks Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Orbis 4K realtime model Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Redemption Agent Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Orbis continuous video Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Avatar Turbo Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 African voice workflow Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 synchronized audio Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 Blender previs Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Creative Agent Canvas Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 single GPU workflow Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Topaz for Web Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Wan 3.0 30-second video Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “WorldGen Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Wan 3.0 image-to-video Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 model comparison Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Max Turbo streaming Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Max 768p pricing Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Grok Imagine 14 references Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Orbis prompt steering Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Max OpenRouter Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Wan 3.0 on SeaArt Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Gemma 4 E4B Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “AI UGC workflow Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Viggle-Animate Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “GWM-2 Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Fable 5.1 Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Topaz Astra 2 Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Lyria 3.5 Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “pet UGC ad Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedream 5 Pro Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “GPT-6 Astra Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Kimi K3 Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax-H3 Semantic-Bridge Adapter Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Equirectangular LoRA Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “30 second continuous vlog Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Fused Turbo Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Product Promo Video Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Clay Renderer Plugin Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Brand system Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Grok Imagine Video 1.5 Agent Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 Fused Turbo INT8 Model identity check”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “Seedance 2.5 Vmake Labs model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 1 entry address model orientation and fit review; representative queries are “MiniMax H3 synced sound model evaluation”. Use that group to identify whether the reader needs provider identity, a model family, an accepted input, a controllable behavior, an output constraint, an access surface, or a production-fit decision. 34 entries address exact-version review; representative queries are “topview ai seedance 2.0”, “veo 3.2”, and “seedance 2.5”. Use that group to treat every numeral, suffix, separator, preview label, and provider identifier as version-sensitive until documentation and the product surface agree. 5 entries address watermark, provenance, and rights review; representative queries are “topyappers sora watermark”, “sora ai watermark”, and “higgsfield watermark”. Use that group to distinguish a visible watermark, embedded metadata, provenance credential, plan-dependent clean export, disclosure label, and unauthorized removal request.

6. 06

   Model query guide: compare evidence, not labels

   Model queries in this directory are not interchangeable products. Before comparing two pages, align the developer, official identifier, release or mode, access surface, region, account terms, evidence date, intended input, required control, output constraint, rights, and delivery target. Then use the same authorized material, brief, attempt allowance, and acceptance criteria. Keep provider documentation, observed behavior, editorial preference, downstream repair, commercial terms, and unknowns in separate columns. A neighboring version, third-party interface, community example, search phrase, or workspace CTA cannot fill a missing field. When a detail page has no dated first-party evidence for the requested fact, its accurate status is Unknown / not verified. The useful outcome is a dated decision for one workflow, not a permanent ranking of every label in the Hub. Record who made the decision, what remains unknown, and when the evidence should be refreshed before reuse.

Availability and capabilities can change; verify in product.