Companies
Browse providers, model families and releases in the catalog.
OpenAI
GPT models for reasoning, software and multimodal workflows.
99 models in the index
Anthropic
The Claude family, from high-volume tasks to long-running agents.
35 models in the index
Gemini language, speech, embedding and robotics models, plus Nano Banana, Veo and Lyria media generation.
120 models in the indexxAI
Grok language, image, video, voice and transcription models, with release-specific capabilities and billing.
25 models in the index
Amazon
Amazon-developed Nova, Titan and reranking models across language, image, video, speech and retrieval.
24 models in the indexMistral AI
A mixed catalog of open-weight and proprietary models.
25 models in the index
WRITER
Palmyra language models for business workflows.
12 models in the index
Upstage
Solar models for document-heavy and multilingual agent workflows.
69 models in the index
Microsoft AI
Microsoft MAI reasoning, coding, image generation, speech synthesis and transcription models, separate from third-party models hosted on Azure.
37 models in the index
Alibaba Cloud
Alibaba develops Qwen, HappyHorse and other language, media and spatial models. Only reviewed closed-weight releases appear here.
140 models in the indexBaidu
Baidu ERNIE language and reasoning models, text embeddings and MuseSteamer image/video services, separated from public weights and third-party Qianfan hosting.
9 models in the indexTencent
Tencent Hunyuan and WAND commercial language, vision, image, video, 3D and speech model services, alongside a separately reviewed public-weight ecosystem.
33 models in the indexByteDance
Seed language and speech, Seedance and Seedream generation, Seed3D assets, avatar and motion models, and Skylark embeddings.
67 models in the index
NAVER Cloud
NAVER Cloud HyperCLOVA X language models, Dolphin embeddings, RAG services, NEST transcription, CSR and multilingual voice synthesis.
12 models in the indexiFLYTEK
Spark language, reasoning and native vision models, including specialist scientific-literature services.
15 models in the index
Cohere
Generative, embedding, reranking, document and speech models with mixed distribution terms.
34 models in the index
AI21 Labs
Jurassic language models, specialized text models and the public-weight Jamba family.
18 models in the indexPerplexity
Search and agent services with multiple underlying models.
18 models in the index
StepFun
StepFun text, vision, speech recognition, voice conversation, speech synthesis and music services, reviewed separately from its public-weight releases.
16 models in the indexSenseTime
SenseNova and SenseChat commercial language, multimodal, realtime speech and licensed edge models, with separate Token Plan and SenseCore access.
22 models in the index
ElevenLabs
Proprietary speech, transcription, music, sound-effects, dubbing and audio-processing models.
28 models in the indexDeepgram
Speech recognition, conversational turn detection and text-to-speech models.
33 models in the indexAssemblyAI
Native speech recognition models for recorded audio and live conversations, with domain-specific transcription add-ons.
9 models in the index
Runway
Video and image generation, character animation, world models and video post-production.
15 models in the index
Black Forest Labs
FLUX image and video models for generation, editing and visual production.
14 models in the index
Ideogram
Image generation, typography and precise editing, with hosted custom-model training.
8 models in the index
Recraft
Proprietary raster and vector image models for generation, visual styles and image editing.
21 models in the indexLuma AI
UNI image models and Ray video models, with generation, reference guidance and media editing.
12 models in the indexMiniMax
Speech, transcription, image and Hailuo video models, language previews, historical music releases and explicitly identified H3 managed components.
31 models in the index
Cartesia
Sonic speech generation and Ink transcription models, with dated snapshots, streaming transports and model-specific voice fine-tuning.
20 models in the index
Hume AI
Octave speech synthesis, EVI voice conversations and expression measurement in audio and faces.
8 models in the index
Sarvam AI
Indian-language speech recognition, speech synthesis, translation and document intelligence, with managed and customer-hosted deployment options.
14 models in the index
Voyage AI
Embedding and reranking models for text, code, documents and visual retrieval.
35 models in the index
Reka
Multimodal language models, unified world-model research and managed video understanding.
9 models in the index
Midjourney
Subscription-based image generation, editing and image-to-video models, including Niji models developed with Spellbrush.
22 models in the index
Adobe
Firefly creative models for images, vectors, layouts, video, speech and audio, plus managed production APIs.
15 models in the index
Kuaishou (Kling)
Kling video, image and voice models, with creative and commerce media APIs from Kuaishou.
33 models in the index
PixVerse
PixVerse develops video generation and interactive world models, alongside managed editing, speech and avatar services.
18 models in the index
Moonvalley
Developer of Marey video models for filmmaking. Moonvalley and Reka announced their merger in June 2026; releases retain their original developer attribution.
2 models in the index
Pika
Developer of Pika video generation, Pika Audio and the PikaStream real-time visual engine. Its hosted third-party model catalog is distinct from its native releases.
16 models in the index
Stability AI
Developer of Stable Audio and managed Stable Image services. Downloadable Stable Diffusion, Stable LM, Stable Code, audio and 3D releases are separated from this closed-weight subset.
20 models in the index
Suno
Developer of hosted music models and Speech, a model for spoken audio with an accompanying soundtrack.
13 models in the indexUdio
Uncharted Labs’ hosted music-generation platform, with v1, v1.5 and v1.5 Allegro access documented during its licensing transition.
3 models in the index
Resemble AI
Developer of speech-conversion, deepfake-detection and media-provenance technology. Its service also integrates models from other developers.
12 models in the index
PlayAI / PlayHT
Developer of PlayHT and PlayDialog speech-generation engines. This collection preserves documented historical releases and their host-specific availability.
6 models in the index
Speechify
Developer of the Simba speech-model family and the SpeechifyAI developer API. Developer API billing is separate from the consumer Reader app.
7 models in the index
Soniox
Developer of native speech-recognition, translation and speech-generation models, with regional API deployments.
16 models in the indexSpeechmatics
Speech-recognition and synthesis developer with general, multilingual, medical and voice-agent models, plus licensed local deployments.
17 models in the index
Fish Audio
Developer of speech synthesis, transcription and voice-design models. The catalog separates proprietary releases from the publicly downloadable Fish Speech, S1-mini and S2 Pro weights.
6 models in the index
Inworld AI
Developer of the Realtime TTS speech families and Inworld STT. Its broader inference platform also routes other developers’ models, which remain attributed to their original creators.
7 models in the index
Gradium
Developer of streaming speech synthesis, recognition and translation, plus Phonon for licensed offline speech generation.
10 models in the index
Rime
Developer of Coda, Mist and Arcana conversational speech models, with cloud APIs, licensed engines and independent partner deployments.
8 models in the index
Murf AI
Developer of proprietary Falcon conversational speech and Gen2 narration models, with voice conversion, translation and dubbing services.
7 models in the index
Rev AI
Speech recognition and transcript intelligence provider, with proprietary API services alongside separately released Reverb research weights.
10 models in the index
DeepL
Specialist language AI for translation, writing, speech and translation-quality evaluation. Its managed API routes requests among proprietary models.
7 models in the index
Descript
Develops proprietary audio and video editing models, with an editor that also integrates third-party speech, image and video models.
6 models in the index
Krisp
Proprietary realtime voice enhancement, accent conversion and conversation-signal models, distributed through local SDKs alongside a cloud translation API.
32 models in the index
Smallest AI
Proprietary Lightning speech synthesis, Pulse transcription, Electron language and Hydra full-duplex speech models, with commercial self-host options.
13 models in the index
Sanas
Proprietary realtime accent conversion, speech restoration, noise suppression and voice-activity models, plus managed multilingual speech translation.
28 models in the index
Zoom
Proprietary speech recognition, conversation summarization, machine translation and specialized small language model research, alongside a federated platform using external models.
7 models in the index
Gladia
Proprietary Solaria speech recognition, historical Whisper-derived ASR and managed audio-intelligence pipelines, with third-party LLM analysis kept under its original creators.
3 models in the index
Modulate
Audio-native Velma models for speech recognition, conversation analysis, source-audio redaction and acoustic detection.
12 models in the index
Reson8
European-language speech recognition from Resonate Labs B.V., with two named Resonant releases and a continuously updated managed API supporting custom terminology and turn detection.
3 models in the index
Sophea AI · KIEFER SA
KIEFER SA’s proprietary speech systems and adaptations, including a hosted ASR K1 preview. Separately released open-weight language, retrieval, speech and robotics families are excluded from this closed-weight catalog.
6 models in the indexMeta
Meta’s Muse API models and research systems for language, coding, images, video, speech and audio/text embeddings. Public-weight Llama, Glimmer, SAM and other released families are excluded.
62 models in the index
Zhipu AI / Z.ai
GLM language and realtime models, CogView and CogVideo generation, embeddings and retrieval services from Zhipu AI.
13 models in the indexMoonshot AI / Kimi
Developer of the Moonshot v1 APIs and Kimi reasoning and research models. Current public-weight Kimi releases are outside this closed-weight directory.
9 models in the index
Inflection AI
Developer of the Inflection foundation models and Pi personal AI. Historical releases and API variants are documented separately from the current Pi product.
6 models in the index
Baichuan AI
Developer of Baichuan general language APIs, domain-trained finance models, retrieval embeddings and medical research systems. Open-weight releases are tracked separately from this directory.
12 models in the index
01.AI / Yi
Developer of Yi language and vision models, now focused on enterprise AI. Public China API access ended September 3, 2026; the international platform also displays a suspension notice.
11 models in the index
360 / Zhinao
Chinese developer of Zhinao language models and speech systems. Its AI platform also routes requests to third-party models, so a Zhinao-branded selector does not always identify a native model.
10 models in the index
Xiaomi / MiMo
Developer of MiMo language, multimodal and speech models. The closed-weight directory includes earlier V2 releases and current TTS variants; the publicly released V2.5/V2.6 language and ASR weights are excluded.
7 models in the index
Samsung
Samsung Research develops the proprietary Gauss family for language, code and images, with later Gauss2 variants designed for on-device and internal cloud workloads.
6 models in the index
Huawei / Pangu
Developer of Pangu models for language, vision, retrieval and industry workloads. Managed ModelArts assets are reviewed separately from the public-weight OpenPangu releases.
57 models in the index
LG AI Research
LG develops EXAONE language and multimodal models. This directory covers its proprietary historical releases and internal research adaptations, while excluding publicly released weights.
13 models in the index
OPPO
OPPO develops AndesGPT for device and cloud workloads and proprietary vision models for ColorOS. Its separate open research models are excluded from this closed-weight directory.
4 models in the index
IBM
IBM’s proprietary Granite, Slate and Watson speech models, with historical releases, API aliases, voice catalogs and public-weight successors kept distinct.
108 models in the index
Apple
Apple develops private models for device and cloud features, coding and image creation. Its third generation includes models co-developed with Google. Public Apple research weights are tracked separately from these product models.
14 models in the index
NTT
NTT develops the proprietary Japanese/English tsuzumi language-model family. NTT DATA and other group companies provide cloud and private deployments with different access and billing terms.
5 models in the index
NEC
NEC develops cotomi language models for Japanese enterprise workloads, with commercial services and private appliance deployments.
6 models in the index
Fujitsu
Fujitsu co-develops Takane with Cohere for Japanese enterprise workloads and private deployments. The lineup includes compact text models and a larger vision model.
7 models in the index
Preferred Networks
Preferred Networks and its group develop the Japanese PLaMo family, spanning commercial language models, edge deployment, finance and translation.
15 models in the index
ELYZA
ELYZA develops Japanese language models, commercial enterprise deployments and specialist research derivatives. Its public releases and privately distributed models have different access terms.
9 models in the index
Ricoh
Ricoh develops Japanese enterprise language models, document-understanding models and private deployments. Its research releases and publicly distributed checkpoints have different access terms.
14 models in the index
ABEJA
ABEJA develops Japanese language models and enterprise AI systems. Its public model releases are distinct from earlier private GPT models and unpublished specialist research checkpoints.
4 models in the index
Maritaca AI
Brazilian developer of Sabiá language models, specializing in Portuguese and Brazilian knowledge, with API access and optional São Paulo processing.
13 models in the index
WideLabs
WideLabs develops proprietary language, vision and speech models for Latin America. Amazônia IA serves Brazilian use cases; PatagonIA was developed with ISCI for Chile.
7 models in the index
World Labs
World Labs develops proprietary spatial models for generating and reconstructing 3D worlds, including Marble, Atlas and RTFM. Its open rendering software is separate from its model weights.
7 models in the index
Vidu (ShengShu)
ShengShu’s Vidu platform offers image, video, audio and real-time character generation. The international API terms identify ShengShu AI HK Limited as its operator.
28 models in the index
fal
fal Research develops post-trained generative-media models, while the fal platform also hosts models from other developers.
2 models in the index
Beatoven.ai
Developer of Maestro for music and sound effects. The founder states that the consumer service ended and proprietary-model licensing continues through Beatoven Private Limited.
1 model in the index
CassetteAI
Pixl Technologies develops CassetteAI music and sound-effect generators, with hosted fal APIs and separate edge-audio development. Its TTS offering is marked coming soon.
2 models in the index
AIVA
AIVA Technologies develops music-generation and audio research models, including Lyra, Influence, OmniCodec and ORAQL. Its music assistant is a separate product surface with subscription limits.
4 models in the index
SOUNDRAW
Japanese developer of a proprietary music engine trained on its in-house catalog. SOUNDRAW offers a music editor and an API for embedding generation into other products.
1 model in the index
Meshy
Developer of 3D generation and texturing models, including Meshy 7.1 and Smart Topology T2. API credit charges and supported selectors vary by operation.
17 models in the index
Hi3D / Hitem3D
Math Magic develops Hi3D, previously branded Hitem3D, for image-based 3D assets, portraits, reliefs and mesh processing. Native API access continues to use api.hitem3d.ai.
12 models in the index
Tripo / VAST
Tripo develops 3D geometry and texture models with text, image and multiview inputs. Its native API terms identify Holymolly as the service company. VAST also releases separate public research models.
8 models in the index
Hyper3D / Deemos
Deemos develops Hyper3D Rodin for 3D generation and processing. Native API access includes three documented generation families plus managed Agentic, Bang and texture workflows.
35 models in the index
Aleph Alpha
German model developer behind Luminous and Pharia. This catalog records its closed-weight Luminous releases; public-weight products are reviewed separately for exclusion.
16 models in the index
Yandex
Developer of Alice AI, YandexGPT, YandexART and SpeechKit. Native models and managed speech services are distinguished from public-weight releases.
44 models in the index
Sber
Developer of GigaChat language and embedding models. Closed hosted models are distinguished from publicly released checkpoints.
17 models in the index
Hedra
Visual model developer and inference platform. Its own Character and Avatar models animate characters from images and supplied audio.
2 models in the index
LemonSlice
Research lab building interactive avatar and character world models. Its native video layer can run with a customer-selected speech and language stack.
9 models in the index
Tavus
Research lab developing human rendering, audiovisual perception, conversational timing and interactive audiovisual models.
13 models in the index
Anam
Developer of proprietary Cara models for live audio-driven face generation, offered alongside configurable speech and language services.
5 models in the index
HeyGen
Developer of avatar rendering and speech models, with identity-based image generation and a MiniMax-derived audiovisual model.
14 models in the index
Sync Labs
Synchronicity Labs develops zero-shot visual lip-sync and performance-editing models, with separate speaker detection and third-party speech integrations.
12 models in the index
D-ID
D-ID develops proprietary visual avatar models for photo animation, recorded presenters and expressive real-time delivery. Connected speech and language services remain separate components.
5 models in the index
Synthesia
Synthesia develops proprietary avatar video and voice models. Its studio, legacy video API and live avatar renderer have different capabilities and access conditions.
13 models in the index
Simli
Simli develops speech-driven visual avatars. Trinity uses a Gaussian representation; its hosted rendering, legacy avatar generation and agent integrations have separate configuration and billing scopes.
4 models in the index
Akool
Akool offers proprietary visual models, multilingual voice engines and hosted avatar, image and video workflows. Several endpoints route to external providers; those integrations are distinguished from independently identified Akool weights.
30 models in the index
Topaz Labs
Image and video enhancement models for upscaling, restoration, denoising, interpolation and color processing.
134 models in the index
Leonardo AI
Proprietary Lucid and Phoenix image models and the Motion video family, with direct and hosted API access.
6 models in the index
Decart
Lucy video and image editing, virtual try-on models, and the Oasis realtime world-model family.
19 models in the index
Higgsfield
Soul image models and DoP video generation, alongside character-reference services and hosted creative workflows.
12 models in the index
Mirage / Captions
Avatar video and expressive voice models from Mirage, the company behind Captions, with a separate managed captioning API.
4 models in the index
Krea
Krea 1 and Krea 3 image models, hosted Krea 2 pipelines, enhancement and realtime creative tools. Downloadable weights and third-party models are distinguished from private models.
7 models in the index
Pruna AI
Pruna operates P-Image and P-Video media models, adapter-training routes and optimized third-party deployments.
13 models in the index
Reve
Reve developed image-generation and editing models using spatial layout representations. Its first-party creation service has ended.
6 models in the index
Photoroom
Photoroom develops image-editing and product-fidelity models and operates managed APIs for commerce imagery.
9 models in the indexLexica
Lexica provides the Aperture image-generation family and an image search service. Public search documentation is separate from its paid web generator.
6 models in the index
Haiper
Haiper develops hosted image and video models. The catalog separates named model identities, version-flexible generation APIs and video enhancement operations.
8 models in the index
Magnific / formerly Freepik
Creative AI platform with Mystic, Magnific One, image/video enhancement workflows and third-party integrations.
14 models in the index
Playground
Developer of Playground image-generation models and the Argus vision-language model. Earlier public-weight releases are excluded from the closed-weight catalog.
2 models in the index
Nixtla
Developer of the closed-source TimeGPT family for forecasting and anomaly detection. The open SDK and Nixtlaverse libraries are separate from TimeGPT weights.
7 models in the index
Nanonets
Document AI provider with OCR research models and configurable extraction workflows. Public OCR weight releases are excluded from private-model counts.
7 models in the index
Snowflake
Enterprise data platform with proprietary Arctic-Extract document models and managed Cortex services. Its public-weight research releases are excluded from closed-weight coverage.
10 models in the index
Cursor
Developer-tool provider training the Composer coding-model family and custom completion models. Hosted third-party models retain their original provider identities.
5 models in the index
Cognition
Developer of SWE software-engineering models, including the Windsurf-origin family, deployed through Devin products and specialized coding tools.
11 models in the index
Poolside
Software-engineering model lab with historical Malibu and Point enterprise models. Current reviewed Laguna families have public weights and are excluded from this closed-weight catalogue.
2 models in the index
Bloomberg
Financial information and technology company that published research on the private BloombergGPT financial language model.
1 model in the index
Magic
Research lab developing long-context and coding models. Recorded LTM models and internal pretraining runs are research, not verified public APIs.
9 models in the index
Tabnine
Coding assistant provider with proprietary chat and completion models, enterprise private deployment and historical Deep TabNine engines.
6 models in the index
Salesforce
CRM provider with private sales, developer and action models alongside public research releases.
7 models in the index
Augment Code
Coding platform with internally trained embedding and completion models, a managed Context Engine and routed Prism ensembles.
8 models in the index
Inception
Developer of proprietary Mercury diffusion language models for reasoning, coding, typed decisions and low-latency agent workflows.
9 models in the index
Relace
Specialized coding-agent models for applying edits, compacting context and retrieving source code, alongside hosted open models.
7 models in the index
Morph
Specialized models for code editing, repository search, context compaction and agent classification, alongside third-party open-weight hosting.
11 models in the index
Contextual AI
Specialized grounded language models, instruction-following retrieval and managed document-understanding components for enterprise RAG.
5 models in the index
Reducto
Private document-parsing and deployment models alongside versioned extraction, OCR and document-workflow components.
27 models in the index
Lakera / Check Point AI Security
Private security model families and managed guardrails for prompt attacks, harmful content, sensitive data and agent behavior.
5 models in the index
Patronus AI
Digital environment simulation and AI evaluation research. Public model releases are distinguished from hosted services whose exact checkpoints remain unverified.
1 model in the index
Hippocratic AI
Healthcare conversation models and the Polaris constellation, with core language, speech and specialist components.
14 models in the index
Nabla
Clinical speech recognition and managed documentation, coding and structured-data workflows.
8 models in the index
Abridge
Clinical speech recognition, documentation and claim verification, combining proprietary models with managed model pipelines.
5 models in the index
Ambience Healthcare
Clinical coding and documentation AI, including a documented private o3-mini derivative and managed clinical workflows.
5 models in the index
SafetyKit
Moderation and risk-analysis models and managed services for content, livestreams and marketplaces.
5 models in the index
Cadence / ChipStack
Chip-design AI with documented private ChipStack RFT derivatives and a separately identified managed agent workflow.
3 models in the index
Runloop
Coding-agent infrastructure provider with a documented private API-code RFT experiment.
1 model in the index
Harvey
Legal and tax AI with custom case-law models, embeddings and private RFT work, alongside managed research-backed products.
6 models in the index
Accordance
Custom tax/accounting model family and a separately identified managed research platform.
2 models in the index
Milo
Historical family-assistant provider with a documented private scheduling RFT experiment.
1 model in the index
Thomson Reuters
Proprietary Thomson models, historical legal custom-model research and separately identified CoCounsel workflows.
5 models in the index
Luminance
Proprietary legal model families and managed contract-intelligence services.
5 models in the index
TypeSafe AI
Jev System One models for typed decisions from text and structured state.
2 models in the index
Liquid AI
Hosted decision-model services and separately published open-weight model families.
2 models in the index
Glean
Enterprise-adapted retrieval and ranking models, private security detection families and managed governance services.
6 models in the index
Lelapa AI
South African provider of proprietary transcription, translation, diarisation and language-identification models in Vulavula.
12 models in the index
Intron
African-accent speech recognition, multilingual speech synthesis and voice authentication through the commercial Sahara family.
11 models in the index
Spitch
African-language speech synthesis, transcription, translation and Yoruba diacritics through managed APIs at spitch.app.
4 models in the index
Pison Labs
African-language speech and language APIs, alongside unreleased named ASR, TTS and embedding research models.
11 models in the index
Abena AI / Mobobi
African-language speech APIs and separately delivered on-device speech SDK components.
5 models in the index
Pindo
African-language speech and entity APIs plus an enterprise phone-agent platform.
4 models in the index
Addis AI
African-language hosted text, speech and realtime APIs alongside public research model releases.
7 models in the index
Namu
Hausa-focused speech translation, transcription, synthesis and voice-conversation families in development.
4 models in the index
9jaLingo
Nigerian-language hosted speech APIs with a public base-voice catalog and separately published open TTS weights.
2 models in the index
NeuraSpeech AI
Airo Global’s managed speech recognition, synthesis and language-translation APIs with subscription plans and custom voice profiles.
5 models in the index
WellSaid
Proprietary voice synthesis with Caruso delivery controls, a Legacy engine and a hosted catalog that includes partner voices.
2 models in the index
Typecast / Neosapience
Proprietary SSFM speech synthesis, expressive voice controls and custom-voice workflows, with separate API and SageMaker offerings.
6 models in the index
CereProc
CereWave neural voices and CereVoice speech/singing services, now presented as part of Capacity. Local deployment products and hosted services have distinct contracts.
3 models in the index
Acapela Group
Proprietary neural speech synthesis, voice preservation and AAC voice technology, with hosted and licensed local deployment options.
7 models in the index
ReadSpeaker
Proprietary neural voice families, licensed local speech packages, hosted speech and custom voice development, including ReadSpeaker Korea. Some hosted voice choices use third-party engines.
8 models in the index
Mubert
Proprietary music composition services using musician-created sounds, with text/image/playlist generation and live streaming. API versions and music libraries are not separately identified neural checkpoints.
2 models in the index
LOVO / Genny
Historically documented Pro speech families, mixed-provider Genny TTS catalogue and custom voice creation. Current public site/app access is unverified; catalogue multilingual voices include ElevenLabs technology.
4 models in the index
Respeecher
Proprietary real-time speech models, Marketplace voice workflows, bespoke voice production and historical audio enhancement. Public Ukrainian Data2Vec weights are excluded from this closed-model catalogue.
10 models in the index
LMNT
Former speech provider behind Aurora and Blizzard, plus managed voice enrollment and conversion workflows. Provider has shut down; entries preserve historical contracts and original branding.
5 models in the index
Altered
Proprietary voice morphing, expressive TTS, audio cleanup and realtime voice augmentation, plus explicitly managed customer-voice workflows. Third-party speech providers in Studio are not relabeled as Altered models.
22 models in the index