[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-together-ai-together-fine-tuning":3,"mdc--9vb1n7-key":34,"related-org-together-ai-together-fine-tuning":612,"related-repo-together-ai-together-fine-tuning":785},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":12,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":29,"sourceUrl":32,"mdContent":33},"together-fine-tuning","fine-tune and adapt models on Together AI","LoRA, full fine-tuning, DPO preference tuning, VLM training, function-calling tuning, reasoning tuning, and BYOM uploads on Together AI. Reach for it whenever the user wants to adapt a model on custom data rather than only run inference, evaluate outputs, or host an existing model.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},"together-ai","Together AI","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Ftogether-ai.jpg","togethercomputer",[13,17,20],{"name":14,"slug":15,"type":16},"LLM","llm","tag",{"name":18,"slug":19,"type":16},"Machine Learning","machine-learning",{"name":21,"slug":22,"type":16},"AI","ai",31,"https:\u002F\u002Fgithub.com\u002Ftogethercomputer\u002Fskills","2026-07-26T05:49:10.243114",null,4,[],{"repoUrl":24,"stars":23,"forks":27,"topics":30,"description":31},[],"Skills to help your coding agents use Together AI products.","https:\u002F\u002Fgithub.com\u002Ftogethercomputer\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Ftogether-fine-tuning","---\nname: together-fine-tuning\ndescription: \"LoRA, full fine-tuning, DPO preference tuning, VLM training, function-calling tuning, reasoning tuning, and BYOM uploads on Together AI. Reach for it whenever the user wants to adapt a model on custom data rather than only run inference, evaluate outputs, or host an existing model.\"\n---\n\n# Together Fine-Tuning\n\n## Overview\n\nUse Together AI fine-tuning when the user needs to adapt a model to their own data or behavior.\n\nSupported workflows in this repo:\n\n- LoRA fine-tuning\n- full fine-tuning\n- DPO preference tuning\n- VLM fine-tuning\n- function-calling fine-tuning\n- reasoning fine-tuning\n- BYOM upload paths\n\n## When This Skill Wins\n\n- Train a model on custom instruction or conversational data\n- Improve function-calling reliability with supervised examples\n- Train on preferences rather than only demonstrations\n- Fine-tune multimodal or reasoning-oriented models\n- Deploy a fine-tuned output model later through dedicated endpoints\n\n## Hand Off To Another Skill\n\n- Use `together-chat-completions` for plain inference without training\n- Use `together-evaluations` to measure a model before or after tuning\n- Use `together-dedicated-model-inference` to host the resulting tuned model\n- Use `together-gpu-clusters` only when the user needs raw infrastructure rather than managed tuning\n\n## Quick Routing\n\n- **Standard LoRA or full fine-tuning**\n  - Start with [scripts\u002Ffinetune_workflow.py](scripts\u002Ffinetune_workflow.py)\n  - Read [references\u002Fdata-formats.md](references\u002Fdata-formats.md)\n- **DPO preference tuning**\n  - Start with [scripts\u002Fdpo_workflow.py](scripts\u002Fdpo_workflow.py)\n- **Function-calling tuning**\n  - Start with [scripts\u002Ffunction_calling_finetune.py](scripts\u002Ffunction_calling_finetune.py)\n- **Reasoning tuning**\n  - Start with [scripts\u002Freasoning_finetune.py](scripts\u002Freasoning_finetune.py)\n- **VLM tuning**\n  - Start with [scripts\u002Fvlm_finetune.py](scripts\u002Fvlm_finetune.py)\n- **Model support and deployment options**\n  - Read [references\u002Fsupported-models.md](references\u002Fsupported-models.md)\n  - Read [references\u002Fdeployment.md](references\u002Fdeployment.md)\n\n## Workflow\n\n1. Choose the tuning method that matches the desired behavior change.\n2. Validate dataset format before spending tokens on training.\n3. Upload training data and keep the returned file ID.\n4. Create the job with explicit method-specific parameters.\n5. Monitor job state, events, checkpoints, and per-step training metrics before handing off to deployment.\n\n## High-Signal Rules\n\n- Python scripts require the Together v2 SDK (`together>=2.0.0`). If the user is on an older version, they must upgrade first: `uv pip install --upgrade \"together>=2.0.0\"`.\n- Prefer LoRA unless the user has a specific reason to pay for full fine-tuning.\n- Keep data-format validation close to the upload step so bad files fail early.\n- Treat deployment as a separate phase; fine-tuning success does not automatically mean serving success.\n- Use the method-specific script instead of overloading one generic workflow for all modes.\n- Parameterize dataset paths, model IDs, and suffixes in automation instead of embedding one demo dataset forever.\n\n## Resource Map\n\n- **Data formats**: [references\u002Fdata-formats.md](references\u002Fdata-formats.md)\n- **Supported models**: [references\u002Fsupported-models.md](references\u002Fsupported-models.md)\n- **Deployment guide**: [references\u002Fdeployment.md](references\u002Fdeployment.md)\n- **LoRA or full workflow**: [scripts\u002Ffinetune_workflow.py](scripts\u002Ffinetune_workflow.py)\n- **DPO workflow**: [scripts\u002Fdpo_workflow.py](scripts\u002Fdpo_workflow.py)\n- **Function-calling workflow**: [scripts\u002Ffunction_calling_finetune.py](scripts\u002Ffunction_calling_finetune.py)\n- **Reasoning workflow**: [scripts\u002Freasoning_finetune.py](scripts\u002Freasoning_finetune.py)\n- **VLM workflow**: [scripts\u002Fvlm_finetune.py](scripts\u002Fvlm_finetune.py)\n\n## Official Docs\n\n- [Fine-tuning Quickstart](https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Ffine-tuning-quickstart)\n- [Data Preparation](https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Ffine-tuning-data-preparation)\n- [Fine-tuning Models](https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Ffine-tuning-models)\n- [Deploying a Fine-Tuned Model](https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Fdeploying-a-fine-tuned-model)\n- [Fine-tuning API](https:\u002F\u002Fdocs.together.ai\u002Freference\u002Fpost-fine-tunes)\n",{"data":35,"body":36},{"name":4,"description":6},{"type":37,"children":38},"root",[39,47,54,60,65,105,111,139,145,198,204,348,354,383,389,438,444,552,558],{"type":40,"tag":41,"props":42,"children":43},"element","h1",{"id":4},[44],{"type":45,"value":46},"text","Together Fine-Tuning",{"type":40,"tag":48,"props":49,"children":51},"h2",{"id":50},"overview",[52],{"type":45,"value":53},"Overview",{"type":40,"tag":55,"props":56,"children":57},"p",{},[58],{"type":45,"value":59},"Use Together AI fine-tuning when the user needs to adapt a model to their own data or behavior.",{"type":40,"tag":55,"props":61,"children":62},{},[63],{"type":45,"value":64},"Supported workflows in this repo:",{"type":40,"tag":66,"props":67,"children":68},"ul",{},[69,75,80,85,90,95,100],{"type":40,"tag":70,"props":71,"children":72},"li",{},[73],{"type":45,"value":74},"LoRA fine-tuning",{"type":40,"tag":70,"props":76,"children":77},{},[78],{"type":45,"value":79},"full fine-tuning",{"type":40,"tag":70,"props":81,"children":82},{},[83],{"type":45,"value":84},"DPO preference tuning",{"type":40,"tag":70,"props":86,"children":87},{},[88],{"type":45,"value":89},"VLM fine-tuning",{"type":40,"tag":70,"props":91,"children":92},{},[93],{"type":45,"value":94},"function-calling fine-tuning",{"type":40,"tag":70,"props":96,"children":97},{},[98],{"type":45,"value":99},"reasoning fine-tuning",{"type":40,"tag":70,"props":101,"children":102},{},[103],{"type":45,"value":104},"BYOM upload paths",{"type":40,"tag":48,"props":106,"children":108},{"id":107},"when-this-skill-wins",[109],{"type":45,"value":110},"When This Skill Wins",{"type":40,"tag":66,"props":112,"children":113},{},[114,119,124,129,134],{"type":40,"tag":70,"props":115,"children":116},{},[117],{"type":45,"value":118},"Train a model on custom instruction or conversational data",{"type":40,"tag":70,"props":120,"children":121},{},[122],{"type":45,"value":123},"Improve function-calling reliability with supervised examples",{"type":40,"tag":70,"props":125,"children":126},{},[127],{"type":45,"value":128},"Train on preferences rather than only demonstrations",{"type":40,"tag":70,"props":130,"children":131},{},[132],{"type":45,"value":133},"Fine-tune multimodal or reasoning-oriented models",{"type":40,"tag":70,"props":135,"children":136},{},[137],{"type":45,"value":138},"Deploy a fine-tuned output model later through dedicated endpoints",{"type":40,"tag":48,"props":140,"children":142},{"id":141},"hand-off-to-another-skill",[143],{"type":45,"value":144},"Hand Off To Another Skill",{"type":40,"tag":66,"props":146,"children":147},{},[148,162,174,186],{"type":40,"tag":70,"props":149,"children":150},{},[151,153,160],{"type":45,"value":152},"Use ",{"type":40,"tag":154,"props":155,"children":157},"code",{"className":156},[],[158],{"type":45,"value":159},"together-chat-completions",{"type":45,"value":161}," for plain inference without training",{"type":40,"tag":70,"props":163,"children":164},{},[165,166,172],{"type":45,"value":152},{"type":40,"tag":154,"props":167,"children":169},{"className":168},[],[170],{"type":45,"value":171},"together-evaluations",{"type":45,"value":173}," to measure a model before or after tuning",{"type":40,"tag":70,"props":175,"children":176},{},[177,178,184],{"type":45,"value":152},{"type":40,"tag":154,"props":179,"children":181},{"className":180},[],[182],{"type":45,"value":183},"together-dedicated-model-inference",{"type":45,"value":185}," to host the resulting tuned model",{"type":40,"tag":70,"props":187,"children":188},{},[189,190,196],{"type":45,"value":152},{"type":40,"tag":154,"props":191,"children":193},{"className":192},[],[194],{"type":45,"value":195},"together-gpu-clusters",{"type":45,"value":197}," only when the user needs raw infrastructure rather than managed tuning",{"type":40,"tag":48,"props":199,"children":201},{"id":200},"quick-routing",[202],{"type":45,"value":203},"Quick Routing",{"type":40,"tag":66,"props":205,"children":206},{},[207,240,259,279,299,319],{"type":40,"tag":70,"props":208,"children":209},{},[210,216],{"type":40,"tag":211,"props":212,"children":213},"strong",{},[214],{"type":45,"value":215},"Standard LoRA or full fine-tuning",{"type":40,"tag":66,"props":217,"children":218},{},[219,230],{"type":40,"tag":70,"props":220,"children":221},{},[222,224],{"type":45,"value":223},"Start with ",{"type":40,"tag":225,"props":226,"children":228},"a",{"href":227},"scripts\u002Ffinetune_workflow.py",[229],{"type":45,"value":227},{"type":40,"tag":70,"props":231,"children":232},{},[233,235],{"type":45,"value":234},"Read ",{"type":40,"tag":225,"props":236,"children":238},{"href":237},"references\u002Fdata-formats.md",[239],{"type":45,"value":237},{"type":40,"tag":70,"props":241,"children":242},{},[243,247],{"type":40,"tag":211,"props":244,"children":245},{},[246],{"type":45,"value":84},{"type":40,"tag":66,"props":248,"children":249},{},[250],{"type":40,"tag":70,"props":251,"children":252},{},[253,254],{"type":45,"value":223},{"type":40,"tag":225,"props":255,"children":257},{"href":256},"scripts\u002Fdpo_workflow.py",[258],{"type":45,"value":256},{"type":40,"tag":70,"props":260,"children":261},{},[262,267],{"type":40,"tag":211,"props":263,"children":264},{},[265],{"type":45,"value":266},"Function-calling tuning",{"type":40,"tag":66,"props":268,"children":269},{},[270],{"type":40,"tag":70,"props":271,"children":272},{},[273,274],{"type":45,"value":223},{"type":40,"tag":225,"props":275,"children":277},{"href":276},"scripts\u002Ffunction_calling_finetune.py",[278],{"type":45,"value":276},{"type":40,"tag":70,"props":280,"children":281},{},[282,287],{"type":40,"tag":211,"props":283,"children":284},{},[285],{"type":45,"value":286},"Reasoning tuning",{"type":40,"tag":66,"props":288,"children":289},{},[290],{"type":40,"tag":70,"props":291,"children":292},{},[293,294],{"type":45,"value":223},{"type":40,"tag":225,"props":295,"children":297},{"href":296},"scripts\u002Freasoning_finetune.py",[298],{"type":45,"value":296},{"type":40,"tag":70,"props":300,"children":301},{},[302,307],{"type":40,"tag":211,"props":303,"children":304},{},[305],{"type":45,"value":306},"VLM tuning",{"type":40,"tag":66,"props":308,"children":309},{},[310],{"type":40,"tag":70,"props":311,"children":312},{},[313,314],{"type":45,"value":223},{"type":40,"tag":225,"props":315,"children":317},{"href":316},"scripts\u002Fvlm_finetune.py",[318],{"type":45,"value":316},{"type":40,"tag":70,"props":320,"children":321},{},[322,327],{"type":40,"tag":211,"props":323,"children":324},{},[325],{"type":45,"value":326},"Model support and deployment options",{"type":40,"tag":66,"props":328,"children":329},{},[330,339],{"type":40,"tag":70,"props":331,"children":332},{},[333,334],{"type":45,"value":234},{"type":40,"tag":225,"props":335,"children":337},{"href":336},"references\u002Fsupported-models.md",[338],{"type":45,"value":336},{"type":40,"tag":70,"props":340,"children":341},{},[342,343],{"type":45,"value":234},{"type":40,"tag":225,"props":344,"children":346},{"href":345},"references\u002Fdeployment.md",[347],{"type":45,"value":345},{"type":40,"tag":48,"props":349,"children":351},{"id":350},"workflow",[352],{"type":45,"value":353},"Workflow",{"type":40,"tag":355,"props":356,"children":357},"ol",{},[358,363,368,373,378],{"type":40,"tag":70,"props":359,"children":360},{},[361],{"type":45,"value":362},"Choose the tuning method that matches the desired behavior change.",{"type":40,"tag":70,"props":364,"children":365},{},[366],{"type":45,"value":367},"Validate dataset format before spending tokens on training.",{"type":40,"tag":70,"props":369,"children":370},{},[371],{"type":45,"value":372},"Upload training data and keep the returned file ID.",{"type":40,"tag":70,"props":374,"children":375},{},[376],{"type":45,"value":377},"Create the job with explicit method-specific parameters.",{"type":40,"tag":70,"props":379,"children":380},{},[381],{"type":45,"value":382},"Monitor job state, events, checkpoints, and per-step training metrics before handing off to deployment.",{"type":40,"tag":48,"props":384,"children":386},{"id":385},"high-signal-rules",[387],{"type":45,"value":388},"High-Signal Rules",{"type":40,"tag":66,"props":390,"children":391},{},[392,413,418,423,428,433],{"type":40,"tag":70,"props":393,"children":394},{},[395,397,403,405,411],{"type":45,"value":396},"Python scripts require the Together v2 SDK (",{"type":40,"tag":154,"props":398,"children":400},{"className":399},[],[401],{"type":45,"value":402},"together>=2.0.0",{"type":45,"value":404},"). If the user is on an older version, they must upgrade first: ",{"type":40,"tag":154,"props":406,"children":408},{"className":407},[],[409],{"type":45,"value":410},"uv pip install --upgrade \"together>=2.0.0\"",{"type":45,"value":412},".",{"type":40,"tag":70,"props":414,"children":415},{},[416],{"type":45,"value":417},"Prefer LoRA unless the user has a specific reason to pay for full fine-tuning.",{"type":40,"tag":70,"props":419,"children":420},{},[421],{"type":45,"value":422},"Keep data-format validation close to the upload step so bad files fail early.",{"type":40,"tag":70,"props":424,"children":425},{},[426],{"type":45,"value":427},"Treat deployment as a separate phase; fine-tuning success does not automatically mean serving success.",{"type":40,"tag":70,"props":429,"children":430},{},[431],{"type":45,"value":432},"Use the method-specific script instead of overloading one generic workflow for all modes.",{"type":40,"tag":70,"props":434,"children":435},{},[436],{"type":45,"value":437},"Parameterize dataset paths, model IDs, and suffixes in automation instead of embedding one demo dataset forever.",{"type":40,"tag":48,"props":439,"children":441},{"id":440},"resource-map",[442],{"type":45,"value":443},"Resource Map",{"type":40,"tag":66,"props":445,"children":446},{},[447,461,474,487,500,513,526,539],{"type":40,"tag":70,"props":448,"children":449},{},[450,455,457],{"type":40,"tag":211,"props":451,"children":452},{},[453],{"type":45,"value":454},"Data formats",{"type":45,"value":456},": ",{"type":40,"tag":225,"props":458,"children":459},{"href":237},[460],{"type":45,"value":237},{"type":40,"tag":70,"props":462,"children":463},{},[464,469,470],{"type":40,"tag":211,"props":465,"children":466},{},[467],{"type":45,"value":468},"Supported models",{"type":45,"value":456},{"type":40,"tag":225,"props":471,"children":472},{"href":336},[473],{"type":45,"value":336},{"type":40,"tag":70,"props":475,"children":476},{},[477,482,483],{"type":40,"tag":211,"props":478,"children":479},{},[480],{"type":45,"value":481},"Deployment guide",{"type":45,"value":456},{"type":40,"tag":225,"props":484,"children":485},{"href":345},[486],{"type":45,"value":345},{"type":40,"tag":70,"props":488,"children":489},{},[490,495,496],{"type":40,"tag":211,"props":491,"children":492},{},[493],{"type":45,"value":494},"LoRA or full workflow",{"type":45,"value":456},{"type":40,"tag":225,"props":497,"children":498},{"href":227},[499],{"type":45,"value":227},{"type":40,"tag":70,"props":501,"children":502},{},[503,508,509],{"type":40,"tag":211,"props":504,"children":505},{},[506],{"type":45,"value":507},"DPO workflow",{"type":45,"value":456},{"type":40,"tag":225,"props":510,"children":511},{"href":256},[512],{"type":45,"value":256},{"type":40,"tag":70,"props":514,"children":515},{},[516,521,522],{"type":40,"tag":211,"props":517,"children":518},{},[519],{"type":45,"value":520},"Function-calling workflow",{"type":45,"value":456},{"type":40,"tag":225,"props":523,"children":524},{"href":276},[525],{"type":45,"value":276},{"type":40,"tag":70,"props":527,"children":528},{},[529,534,535],{"type":40,"tag":211,"props":530,"children":531},{},[532],{"type":45,"value":533},"Reasoning workflow",{"type":45,"value":456},{"type":40,"tag":225,"props":536,"children":537},{"href":296},[538],{"type":45,"value":296},{"type":40,"tag":70,"props":540,"children":541},{},[542,547,548],{"type":40,"tag":211,"props":543,"children":544},{},[545],{"type":45,"value":546},"VLM workflow",{"type":45,"value":456},{"type":40,"tag":225,"props":549,"children":550},{"href":316},[551],{"type":45,"value":316},{"type":40,"tag":48,"props":553,"children":555},{"id":554},"official-docs",[556],{"type":45,"value":557},"Official Docs",{"type":40,"tag":66,"props":559,"children":560},{},[561,572,582,592,602],{"type":40,"tag":70,"props":562,"children":563},{},[564],{"type":40,"tag":225,"props":565,"children":569},{"href":566,"rel":567},"https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Ffine-tuning-quickstart",[568],"nofollow",[570],{"type":45,"value":571},"Fine-tuning Quickstart",{"type":40,"tag":70,"props":573,"children":574},{},[575],{"type":40,"tag":225,"props":576,"children":579},{"href":577,"rel":578},"https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Ffine-tuning-data-preparation",[568],[580],{"type":45,"value":581},"Data Preparation",{"type":40,"tag":70,"props":583,"children":584},{},[585],{"type":40,"tag":225,"props":586,"children":589},{"href":587,"rel":588},"https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Ffine-tuning-models",[568],[590],{"type":45,"value":591},"Fine-tuning Models",{"type":40,"tag":70,"props":593,"children":594},{},[595],{"type":40,"tag":225,"props":596,"children":599},{"href":597,"rel":598},"https:\u002F\u002Fdocs.together.ai\u002Fdocs\u002Fdeploying-a-fine-tuned-model",[568],[600],{"type":45,"value":601},"Deploying a Fine-Tuned Model",{"type":40,"tag":70,"props":603,"children":604},{},[605],{"type":40,"tag":225,"props":606,"children":609},{"href":607,"rel":608},"https:\u002F\u002Fdocs.together.ai\u002Freference\u002Fpost-fine-tunes",[568],[610],{"type":45,"value":611},"Fine-tuning API",{"items":613,"total":784},[614,635,647,666,680,691,708,717,732,745,751,765],{"slug":615,"name":615,"fn":616,"description":617,"org":618,"tags":619,"stars":632,"repoUrl":633,"updatedAt":634},"blog-post","write structured blog posts","Write structured long-form blog posts with a consistent structure and SEO optimization.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[620,623,626,629],{"name":621,"slug":622,"type":16},"Content Creation","content-creation",{"name":624,"slug":625,"type":16},"Marketing","marketing",{"name":627,"slug":628,"type":16},"SEO","seo",{"name":630,"slug":631,"type":16},"Writing","writing",1151,"https:\u002F\u002Fgithub.com\u002Ftogethercomputer\u002Ftogether-cookbook","2026-07-17T06:07:39.475433",{"slug":636,"name":636,"fn":637,"description":638,"org":639,"tags":640,"stars":632,"repoUrl":633,"updatedAt":646},"social-media","create social media content","Create social media content including Twitter\u002FX threads, LinkedIn posts, and short-form updates.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[641,642,643,645],{"name":621,"slug":622,"type":16},{"name":624,"slug":625,"type":16},{"name":644,"slug":636,"type":16},"Social Media",{"name":630,"slug":631,"type":16},"2026-07-17T06:07:38.455058",{"slug":648,"name":648,"fn":649,"description":650,"org":651,"tags":652,"stars":23,"repoUrl":24,"updatedAt":665},"together-audio","process audio with Together AI","Text-to-speech and speech-to-text via Together AI, including REST, streaming, and realtime WebSocket TTS, plus transcription, translation, diarization, timestamps, and live STT. Reach for it whenever the user needs audio in or audio out on Together AI rather than chat generation, image or video creation, or model training.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[653,656,659,662],{"name":654,"slug":655,"type":16},"Audio","audio",{"name":657,"slug":658,"type":16},"Speech","speech",{"name":660,"slug":661,"type":16},"Text-to-Speech","text-to-speech",{"name":663,"slug":664,"type":16},"Transcription","transcription","2026-07-26T05:49:08.246858",{"slug":667,"name":667,"fn":668,"description":669,"org":670,"tags":671,"stars":23,"repoUrl":24,"updatedAt":679},"together-batch-inference","run asynchronous batch inference jobs","High-volume, asynchronous offline inference at up to 50% lower cost via Together AI's Batch API. Prepare JSONL inputs, upload files, create jobs, poll status, and download outputs. Reach for it whenever the user needs non-interactive bulk inference rather than real-time chat or evaluation jobs.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[672,675,676],{"name":673,"slug":674,"type":16},"Automation","automation",{"name":14,"slug":15,"type":16},{"name":677,"slug":678,"type":16},"Performance","performance","2026-07-17T06:08:23.919602",{"slug":159,"name":159,"fn":681,"description":682,"org":683,"tags":684,"stars":23,"repoUrl":24,"updatedAt":690},"generate text with Together AI","Real-time and streaming text generation via Together AI's OpenAI-compatible chat\u002Fcompletions API, including multi-turn conversations, tool and function calling, structured JSON outputs, and reasoning models. Reach for it whenever the user wants to build or debug text generation on Together AI, unless they specifically need batch jobs, embeddings, fine-tuning, dedicated endpoints, dedicated containers, or GPU clusters.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[685,686,689],{"name":21,"slug":22,"type":16},{"name":687,"slug":688,"type":16},"API Development","api-development",{"name":14,"slug":15,"type":16},"2026-07-26T05:49:11.244104",{"slug":692,"name":692,"fn":693,"description":694,"org":695,"tags":696,"stars":23,"repoUrl":24,"updatedAt":707},"together-dedicated-containers","deploy custom inference containers on Together AI","Custom Dockerized inference workers on Together AI's managed GPU infrastructure. Build with Sprocket SDK, configure with Jig CLI, submit async queue jobs, and poll results. Reach for it whenever the user needs container-level control rather than a standard model endpoint or raw cluster.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[697,700,703,706],{"name":698,"slug":699,"type":16},"AI Infrastructure","ai-infrastructure",{"name":701,"slug":702,"type":16},"Deployment","deployment",{"name":704,"slug":705,"type":16},"Docker","docker",{"name":18,"slug":19,"type":16},"2026-07-26T05:49:09.267892",{"slug":183,"name":183,"fn":709,"description":710,"org":711,"tags":712,"stars":23,"repoUrl":24,"updatedAt":716},"deploy and operate Together AI models","Deploy and operate models on dedicated GPUs with Together AI's Dedicated Model Inference (DMI, the v2 dedicated endpoints API): beta endpoints, deployments, deployment profiles and hardware configs, autoscaling, traffic splitting, A\u002FB tests, shadow experiments, Prometheus metrics, and custom model or LoRA adapter uploads. Reach for it whenever the user mentions together beta endpoints or tg beta commands, client.beta.endpoints, DMI resources like ep_\u002Fdep_\u002Fcr_\u002Fml_ IDs, or wants production model serving with traffic management on Together AI. This is the current dedicated-hosting API and also covers migrating off the retired legacy v1 endpoints API (non-beta client.endpoints \u002F together endpoints), whose create and restart now return HTTP 403.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[713,714,715],{"name":698,"slug":699,"type":16},{"name":701,"slug":702,"type":16},{"name":14,"slug":15,"type":16},"2026-07-26T06:08:44.044039",{"slug":718,"name":718,"fn":719,"description":720,"org":721,"tags":722,"stars":23,"repoUrl":24,"updatedAt":731},"together-embeddings","generate embeddings and build RAG pipelines","Dense vector embeddings, semantic search, RAG pipelines, and reranking via Together AI. Generate embeddings with open-source models and rerank results behind dedicated endpoints. Reach for it whenever the user needs vector representations or retrieval quality improvements rather than direct text generation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[723,724,725,728],{"name":21,"slug":22,"type":16},{"name":14,"slug":15,"type":16},{"name":726,"slug":727,"type":16},"RAG","rag",{"name":729,"slug":730,"type":16},"Search","search","2026-07-26T05:49:06.247906",{"slug":171,"name":171,"fn":733,"description":734,"org":735,"tags":736,"stars":23,"repoUrl":24,"updatedAt":744},"evaluate LLM outputs with Together AI","LLM-as-a-judge evaluation framework on Together AI. Classify, score, and compare model outputs, select judge models, use external-provider judges or targets, poll results and download reports. Reach for it whenever the user wants to benchmark outputs, grade responses, compare A\u002FB variants, or operationalize automated evaluations.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[737,740,743],{"name":738,"slug":739,"type":16},"Benchmarking","benchmarking",{"name":741,"slug":742,"type":16},"Evals","evals",{"name":14,"slug":15,"type":16},"2026-07-26T05:49:07.241553",{"slug":4,"name":4,"fn":5,"description":6,"org":746,"tags":747,"stars":23,"repoUrl":24,"updatedAt":25},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[748,749,750],{"name":21,"slug":22,"type":16},{"name":14,"slug":15,"type":16},{"name":18,"slug":19,"type":16},{"slug":195,"name":195,"fn":752,"description":753,"org":754,"tags":755,"stars":23,"repoUrl":24,"updatedAt":764},"orchestrate GPU clusters on Together AI","On-demand and reserved GPU clusters (H100, H200, B200) on Together AI with Kubernetes or Slurm orchestration, shared storage, credential management, and cluster scaling for ML and HPC jobs. Reach for it when the user needs multi-node compute or infrastructure control rather than a managed model endpoint.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[756,757,760,763],{"name":698,"slug":699,"type":16},{"name":758,"slug":759,"type":16},"Cloud","cloud",{"name":761,"slug":762,"type":16},"Kubernetes","kubernetes",{"name":18,"slug":19,"type":16},"2026-07-26T05:49:05.252646",{"slug":766,"name":766,"fn":767,"description":768,"org":769,"tags":770,"stars":23,"repoUrl":24,"updatedAt":783},"together-images","generate and edit images with Together AI","Text-to-image generation and image editing via Together AI, including FLUX and Kontext models, LoRA-based styling, reference-image guidance, and local image downloads. Reach for it whenever the user wants to generate or edit images on Together AI rather than create videos or build text-only chat applications.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[771,774,777,780],{"name":772,"slug":773,"type":16},"Creative","creative",{"name":775,"slug":776,"type":16},"Design","design",{"name":778,"slug":779,"type":16},"Image Generation","image-generation",{"name":781,"slug":782,"type":16},"Multimodal","multimodal","2026-07-17T06:04:23.067755",16,{"items":786,"total":832},[787,794,800,806,813,819,826],{"slug":648,"name":648,"fn":649,"description":650,"org":788,"tags":789,"stars":23,"repoUrl":24,"updatedAt":665},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[790,791,792,793],{"name":654,"slug":655,"type":16},{"name":657,"slug":658,"type":16},{"name":660,"slug":661,"type":16},{"name":663,"slug":664,"type":16},{"slug":667,"name":667,"fn":668,"description":669,"org":795,"tags":796,"stars":23,"repoUrl":24,"updatedAt":679},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[797,798,799],{"name":673,"slug":674,"type":16},{"name":14,"slug":15,"type":16},{"name":677,"slug":678,"type":16},{"slug":159,"name":159,"fn":681,"description":682,"org":801,"tags":802,"stars":23,"repoUrl":24,"updatedAt":690},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[803,804,805],{"name":21,"slug":22,"type":16},{"name":687,"slug":688,"type":16},{"name":14,"slug":15,"type":16},{"slug":692,"name":692,"fn":693,"description":694,"org":807,"tags":808,"stars":23,"repoUrl":24,"updatedAt":707},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[809,810,811,812],{"name":698,"slug":699,"type":16},{"name":701,"slug":702,"type":16},{"name":704,"slug":705,"type":16},{"name":18,"slug":19,"type":16},{"slug":183,"name":183,"fn":709,"description":710,"org":814,"tags":815,"stars":23,"repoUrl":24,"updatedAt":716},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[816,817,818],{"name":698,"slug":699,"type":16},{"name":701,"slug":702,"type":16},{"name":14,"slug":15,"type":16},{"slug":718,"name":718,"fn":719,"description":720,"org":820,"tags":821,"stars":23,"repoUrl":24,"updatedAt":731},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[822,823,824,825],{"name":21,"slug":22,"type":16},{"name":14,"slug":15,"type":16},{"name":726,"slug":727,"type":16},{"name":729,"slug":730,"type":16},{"slug":171,"name":171,"fn":733,"description":734,"org":827,"tags":828,"stars":23,"repoUrl":24,"updatedAt":744},{"slug":8,"name":9,"logoUrl":10,"githubOrg":11},[829,830,831],{"name":738,"slug":739,"type":16},{"name":741,"slug":742,"type":16},{"name":14,"slug":15,"type":16},14]