Map / Core AI / Foundation Models
Foundation Models
107 tools+ 1 communityThe AI model families themselves: language models like Claude, GPT, Gemini, Qwen, and DeepSeek, coding models like Codex and Devstral, plus embedding, image, video, and speech model weights. Reasoning is now a mode inside each family, not a family of its own. One card per family; versions and tiers live inside each card, and the products built on these models live in their own categories.
- Solid: curated by our editors.
- Dashed: added by makers, not reviewed yet.
- Seal: Verified maker.
Language Models 40
ChatGPT ChatGPTOpenAI's flagship family, unifying the former o-series reasoners3 use thisNo ratings yetTap again to open
ClaudeClaudeAnthropic's model family in Haiku, Sonnet, Opus, and Fable tiers2 use this5.0 (1)5.0 out of 5, 1 ratingsTap again to open 
GeminiGeminiGoogle DeepMind's multimodal family with built-in thinking modes1 use thisNo ratings yetTap again to open 
GrokGrokxAI's frontier model family with real-time X data access1 use thisNo ratings yetTap again to open 
MistralMistralMistral AI's model family, from Small and Medium up to Large1 use thisNo ratings yetTap again to open 
Amazon NovaAmazon NovaAmazon's multimodal model family served through Bedrock0 use thisNo ratings yetTap again to open 
AmpersandAmpersandCollaborator model for helping groups reach shared goals, via API0 use thisNo ratings yetTap again to open 
BeamBeamReflection AI's 501B open-weight model for coding and agent work0 use thisNo ratings yetTap again to open 
CommandCommandCohere's RAG and tool-use optimized enterprise model family0 use thisNo ratings yetTap again to open 
DeepSeek-V DeepSeek-VDeepSeek's MIT-licensed MoE family with hybrid reasoning (V4, R1)0 use thisNo ratings yetTap again to open
FalconFalconTII's open-weights language model family0 use thisNo ratings yetTap again to open 
GemmaGemmaGoogle's lightweight open-weights model family0 use thisNo ratings yetTap again to open 
GLMGLMZ.ai's open-weights MoE family for agentic coding and reasoning0 use thisNo ratings yetTap again to open - gpt-ossgpt-ossOpenAI's Apache 2.0 open-weights reasoning models (20B and 120B)0 use thisNo ratings yetTap again to open
GraniteGraniteIBM's Apache 2.0 enterprise model family for language and code0 use thisNo ratings yetTap again to open 
InklingInklingOpen-weight multimodal MoE model from Thinking Machines Lab0 use thisNo ratings yetTap again to open 
JambaJambaAI21's hybrid Mamba-Transformer open models for long context0 use thisNo ratings yetTap again to open 
KimiKimiMoonshot AI's trillion-scale open-weights MoE family for agents0 use thisNo ratings yetTap again to open 
LFMLFMLiquid AI's open-weights models built to run on phones, laptops and cars0 use thisNo ratings yetTap again to open 
LlamaLlamaMeta's open-weights model family, succeeded by Muse0 use thisNo ratings yetTap again to open 
MiMo MiMoXiaomi's open-weights reasoning and agentic model family0 use thisNo ratings yetTap again to open
MiniMax-M MiniMax-MMiniMax's open-weights MoE family for long-context agentic work0 use thisNo ratings yetTap again to open- Mobile
LLM MobileLLMMeta's sub-billion parameter models for mobile devices0 use thisNo ratings yetTap again to open - MuseMuseMeta Superintelligence Labs' model family, incl. open-weight Muse Glimmer0 use thisNo ratings yetTap again to open
NanbeigeNanbeigeBOSS Zhipin's open-weights family of small agentic models, from 3B to 33B0 use thisNo ratings yetTap again to open 
NemotronNemotronNVIDIA's open model family with open data and training recipes0 use thisNo ratings yetTap again to open 
OLMoOLMoAi2's fully open models with released data, code, and weights0 use thisNo ratings yetTap again to open 
PhiPhiMicrosoft's small but capable open-weights model family0 use thisNo ratings yetTap again to open 
Pokee IsaacPokee IsaacPokee AI's 28B agentic model preview with a 10M-token context window0 use thisNo ratings yetTap again to open 
QwenQwenAlibaba's open-weights family, from small dense to trillion-scale MoE0 use thisNo ratings yetTap again to open 
Sakana FuguSakana FuguModel that coordinates a pool of LLMs behind one API0 use thisNo ratings yetTap again to open - Sakana NamazuSakana NamazuJapanese-tuned LLM with web search and code execution0 use thisNo ratings yetTap again to open
Sarvam 105BSarvam 105BOpen 105B MoE reasoning model trained in India for Indian languages0 use thisNo ratings yetTap again to open - Sarvam 30BSarvam 30BOpen 30B MoE model for real-time conversation in Indian languages0 use thisNo ratings yetTap again to open
SmolLM SmolLMHugging Face's tiny yet capable fully open language models0 use thisNo ratings yetTap again to open
SolarSolarUpstage's model family, from open-weights Solar Open to the agentic Solar Pro0 use thisNo ratings yetTap again to open 
StepStepStepFun's open-weights MoE family for multimodal agentic reasoning0 use thisNo ratings yetTap again to open 
TrinityTrinityArcee AI's open-weight MoE models from on-device to 400B0 use thisNo ratings yetTap again to open 
Zamba2Zamba2Hybrid Mamba2-transformer small language models0 use thisNo ratings yetTap again to open - ZAYA1ZAYA1Open MoE language models trained end to end on AMD0 use thisNo ratings yetTap again to open
Coding Models 14
CoDA CoDASalesforce AI Research's diffusion language model for code0 use thisNo ratings yetTap again to open- Code World ModelCode World ModelMeta's LLM trained to reason about how code changes program state0 use thisNo ratings yetTap again to open
Cognition SWECognition SWECognition's in-house agentic coding models behind Devin0 use thisNo ratings yetTap again to open 
ComposerComposerCursor's in-house agent model, built for fast multi-step edits0 use thisNo ratings yetTap again to open 
DevstralDevstralMistral AI's open-weights code agent models in 24B and 123B sizes0 use thisNo ratings yetTap again to open 
Doubao-Seed-CodeDoubao-Seed-CodeByteDance's coding model on Volcengine, integrated into Trae IDE0 use thisNo ratings yetTap again to open 
FernFernMirendil's long-context model for agentic coding and AI research work0 use thisNo ratings yetTap again to open 
FlintFlintOpen 40B MoE coding model built to run locally on laptops0 use thisNo ratings yetTap again to open 
Inception MercuryInception MercuryInception Labs' diffusion LLM family for fast chat, code, and edits0 use thisNo ratings yetTap again to open - KAT-DevKAT-DevKwaipilot's open-weights software engineering model family0 use thisNo ratings yetTap again to open
LagunaLagunaPoolside's open-weight MoE models for agentic coding0 use thisNo ratings yetTap again to open 
MellumMellumJetBrains' Apache 2.0 open-weights models for code completion0 use thisNo ratings yetTap again to open - Qwen CoderQwen CoderAlibaba's agentic code models, up to 480B MoE with 35B active0 use thisNo ratings yetTap again to open
- RnjRnjEssential AI's model built from scratch for code and STEM0 use thisNo ratings yetTap again to open
Embedding Models 16
ALIGNALIGNGoogle's large-scale visual-language embedding model0 use thisNo ratings yetTap again to open - BGE-M3BGE-M3BAAI's multilingual dense, sparse, and multi-vector embeddings0 use thisNo ratings yetTap again to open
- CLIPCLIPOpenAI's contrastive image-text embedding model0 use thisNo ratings yetTap again to open
- Cohere EmbedCohere EmbedCohere's multilingual, multimodal embedding models0 use thisNo ratings yetTap again to open
- E5E5Microsoft's instruction-tuned text embedding model family0 use thisNo ratings yetTap again to open
- Embedding
Gemma EmbeddingGemmaGoogle's on-device multilingual embedding models built on Gemma0 use thisNo ratings yetTap again to open - Gemini EmbeddingGemini EmbeddingGoogle's API embedding models, near the top of MTEB0 use thisNo ratings yetTap again to open
- Image
Bind ImageBindMeta's unified embedding across 6 modalities0 use thisNo ratings yetTap again to open 
Jina EmbeddingsJina EmbeddingsLong-context multilingual text and multimodal embeddings0 use thisNo ratings yetTap again to open - LFM2.5 EmbeddingLFM2.5 EmbeddingSmall multilingual embedding and ColBERT retrieval models0 use thisNo ratings yetTap again to open
Nomic EmbedNomic EmbedNomic's fully open text and code embeddings with open training data0 use thisNo ratings yetTap again to open - NV-EmbedNV-EmbedNVIDIA's LLM-based embeddings, non-commercial license0 use thisNo ratings yetTap again to open
- Open
AI Embeddings OpenAI EmbeddingsOpenAI's text-embedding-3 models with shortenable dimensions0 use thisNo ratings yetTap again to open - Qwen EmbeddingQwen EmbeddingAlibaba's Apache 2.0 multilingual embedding and reranking models0 use thisNo ratings yetTap again to open
- Sig
LIP SigLIPGoogle's sigmoid-loss image-text embedding models0 use thisNo ratings yetTap again to open 
Voyage AIVoyage AIMongoDB-owned retrieval embeddings with a shared space across sizes0 use thisNo ratings yetTap again to open
Image & Video Models 12+ 1 community
FluxFluxBlack Forest Labs' image generation and editing model family0 use thisNo ratings yetTap again to open - GPT ImageGPT ImageOpenAI's image generation and editing models, successor to DALL-E0 use thisNo ratings yetTap again to open
HunyuanVideo HunyuanVideoTencent's open-weights text and image to video model0 use thisNo ratings yetTap again to open
ImagenImagenGoogle DeepMind's text-to-image model family0 use thisNo ratings yetTap again to open 
Nano BananaNano BananaGemini's native image generation and conversational editing model0 use thisNo ratings yetTap again to open - Qwen-ImageQwen-ImageAlibaba's open-weights image model with strong text rendering0 use thisNo ratings yetTap again to open
- SeedanceSeedanceByteDance Seed's multi-shot cinematic video generation model0 use thisNo ratings yetTap again to open
- SeedreamSeedreamByteDance Seed's 4K image generation and editing model0 use thisNo ratings yetTap again to open
Stable DiffusionStable DiffusionStability AI's open-weights image generation model family0 use thisNo ratings yetTap again to open 
Uni-1Uni-1Luma AI's unified image generation and editing model0 use thisNo ratings yetTap again to open 
VeoVeoGoogle DeepMind's video model with native synchronized audio0 use thisNo ratings yetTap again to open 
WanWanAlibaba's open-weights MoE video generation model family0 use thisNo ratings yetTap again to open
Speech & Audio Models 15
BarkBarkSuno's early open text-to-audio model, no longer actively updated0 use thisNo ratings yetTap again to open 
ChatterboxChatterboxResemble AI's MIT-licensed TTS with emotion control and cloning0 use thisNo ratings yetTap again to open - Cosy
Voice CosyVoiceAlibaba's multilingual streaming TTS with zero-shot voice cloning0 use thisNo ratings yetTap again to open 
DiaDiaNari Labs' open dialogue TTS with nonverbal cues like laughter0 use thisNo ratings yetTap again to open 
Fish SpeechFish SpeechFish Audio's multilingual TTS and voice cloning model0 use thisNo ratings yetTap again to open 
hertz-devhertz-devOpen base model for full-duplex conversational speech0 use thisNo ratings yetTap again to open 
HibikiHibikiReal-time speech-to-speech translation that keeps the speaker's voice0 use thisNo ratings yetTap again to open 
IndexTTS IndexTTSBilibili's zero-shot TTS with emotion and duration control0 use thisNo ratings yetTap again to open- KokoroKokoroLightweight 82M Apache 2.0 TTS model with high quality output0 use thisNo ratings yetTap again to open
- LFM2.5-AudioLFM2.5-Audio1.5B end-to-end speech and text model for real-time voice0 use thisNo ratings yetTap again to open
- MoshiMoshiKyutai's open full-duplex speech model for real-time voice dialogue0 use thisNo ratings yetTap again to open
- Music
Gen MusicGenMeta's controllable text-to-music generation model0 use thisNo ratings yetTap again to open - Pocket TTSPocket TTS100M-parameter text-to-speech that runs in real time on a CPU0 use thisNo ratings yetTap again to open
Sesame CSMSesame CSMSesame's conversational speech model for natural voice dialogue0 use thisNo ratings yetTap again to open 
XTTSXTTSCoqui's multilingual voice cloning TTS, community-maintained fork0 use thisNo ratings yetTap again to open
Science & Biology Models 6
AlphaFold 3 AlphaFold 3Google DeepMind's biomolecular structure prediction model, weights on request0 use thisNo ratings yetTap again to open
Chai-1Chai-1Chai Discovery's open-weights biomolecular structure prediction model0 use thisNo ratings yetTap again to open 
ESMESMBiohub's protein language and structure prediction model family0 use thisNo ratings yetTap again to open 
ether0ether0Open 24B reasoning model post-trained for chemistry and molecule design0 use thisNo ratings yetTap again to open - GlyphGlyphOpen models that turn chemical structure images into machine-readable molecules0 use thisNo ratings yetTap again to open
- ZUNAZUNAOpen EEG foundation model for denoising and reconstructing brain signals0 use thisNo ratings yetTap again to open
Decision Models 4
ClefClefCloudflare's open-weight decision models, hosted on Workers AI0 use thisNo ratings yetTap again to open - d1d1Decision models that return calibrated answer probabilities in one pass0 use thisNo ratings yetTap again to open
JevJevTypeSafe's decision model: typed questions in, calibrated probabilities out0 use thisNo ratings yetTap again to open 
Perplexity DeciderPerplexity DeciderOpen-weight decision models that return probabilities instead of text0 use thisNo ratings yetTap again to open
Built something for Foundation Models? Submit it and it appears here with a dashed edge. When an editor adds it to the map, it gets the solid one.
Submit a tool