Language Model Deployment and Fine-tuning

4/15/2026

This post introduces Seedance 2.0, a new video generation model on Replicate. It details Seedance 2.0's capabilities, including its ability to reference multiple input types (images, video, audio, text) for greater control over video generation. It highlights features like character consistency, motion transfer, style and composition control, and audio-driven rhythm. The post also emphasizes the model's unified architecture for synchronized audio and video, its handling of physics for realistic motion, and its multi-shot output with camera planning. Examples showcase its use for creating complex scenes, character animations, and physics-based simulations.

11/26/2025

Introduced Perceptron AI's Isaac 0.1, a 2B-parameter vision-language model, to the Replicate platform. Highlighted its capabilities in grounded visual reasoning, strong OCR in real-world conditions, spatial awareness, and efficiency for real-time or edge-constrained applications. Provided API examples for JavaScript.

10/2/2025

This post introduces IBM's Granite 4.0 family of open-source small language models to Replicate. It details their hybrid architecture (Mamba-2 + Transformers) and MoE routing strategy, which enables efficient inference on consumer GPUs for tasks like document summarization, RAG, and AI agents. API usage examples for cURL, JavaScript, and Python are provided, along with a discussion of the models' practical applications and open-source licensing.

8/1/2025

This post details the integration of image input capabilities into the Veo 3 model on Replicate. It explains how to leverage image inputs for style preservation, maintaining artistic integrity from cartoons to photographs. The post also covers Veo 3's proficiency in handling typography within animations and demonstrates how image inputs provide greater creative control over video generation by allowing users to pre-define aesthetics before animating. Specific examples showcase animating parts of an image, generating hyperrealistic footage, and creative prompting techniques.

5/22/2025

This post announces the integration of OpenAI's latest models, including GPT-4.1, GPT-4o, and the o-series, onto the Replicate platform. It details the capabilities of these new models, such as long context handling, multimodal understanding, and structured reasoning. The post also provides a code example demonstrating how to run GPT-4.1 using Replicate's JavaScript client and highlights the availability of OpenAI's image models (GPT-image-1, DALL-E 3) and an audio transcription model (GPT-4o-transcribe).

12/16/2024

This post details the emergence of advanced AI video generation models, highlighting several specific models available on Replicate. It provides a comparative analysis of these models based on ELO score, speed, duration, resolution, and open-source availability. It also mentions the fine-tuning capabilities for models like Genmo Mochi 1 and the ongoing development of new models and features.

7/23/2024

This post details the integration and API access for Meta's Llama 3.1 405B model on Replicate. It provides code examples for running the model via JavaScript, Python, and cURL, demonstrating how to use the `replicate.stream` function for real-time output. It also outlines the model's technical specifications (405 billion parameters, 8000 token context window, multilingual support) and links to relevant resources for responsible AI development and a demo chat app.

7/12/2024

This post contributes to the language model deployment and fine-tuning thread by highlighting new models and tools, and by discussing advancements in data curation, synthetic data generation, custom classifier creation, and function-calling dataset generation. It also mentions new text-to-image models and novel ways to embed LLMs.

4/23/2024

This post details how to run the Snowflake Arctic language model via the Replicate API. It provides code examples for JavaScript, Python, and cURL, demonstrating how to integrate Arctic into applications. It also highlights the API playground for interactive testing and links to relevant documentation and client libraries.

4/18/2024

This post introduces the integration and API access for Meta's Llama 3 models (8B and 70B parameter, base and instruct-tuned variants) on the Replicate platform. It provides concrete code examples for running these models using JavaScript, Python, and cURL, demonstrating API calls, input parameters, and streaming output. It also details the different Llama 3 variants available and links to relevant documentation and a demo application.

1/30/2024

This post introduces the availability of Code Llama 70B on Replicate, detailing its three variants (Base, Python, Instruct) and providing concrete code examples for running it via API using JavaScript, Python, and cURL. It specifies the model version hash (`a279116fe47a0f65701a8817188601e2fe8f4b9e04a518789655ea7b995851bf`) and demonstrates how to set API tokens and construct prediction requests with relevant input parameters for code generation.

12/6/2023

This post details the technical implementation of an AI narrator system by integrating three distinct AI model types: a vision model for image understanding, a language model for script generation, and a text-to-speech model for narration. It provides specific model recommendations (Llava 13B, GPT-4-Vision, Mistral 7B, ElevenLabs, XTTS-v2), discusses input/output formats, latency considerations, and provides code snippets for webcam image capture and model interaction. The post also highlights the ability to combine vision and language tasks into a single LLM prediction.

10/9/2023

Introduced and demonstrated the use of Llama 2 models with built-in grammar support for constrained text generation, specifically for information extraction tasks. Showcased how to use JSON schemas to guide LLM output into structured formats, providing a practical example of parsing flight confirmation emails into a JSON object using Replicate's JavaScript client library and a grammar-enabled Llama 2 model.

10/6/2023

This post introduces the Mistral 7B and Mistral 7B Instruct models to Replicate, detailing their technical advantages (recent training data, grouped-query attention, sliding window attention) and providing concrete API examples for running them using JavaScript, Python, and cURL. It also links to existing fine-tuned versions and comparison tools.

8/14/2023

Introduced server-sent event streams for language models, allowing for real-time token-by-token output. Provided Node.js and cURL examples for consuming these streams. Documented how to add streaming support to custom Cog models by yielding tokens.

7/20/2023

This post details the process of fine-tuning Llama 2 models on Replicate. It covers selecting appropriate Llama 2 base or chat models, preparing training data in JSONL format, creating a new model on Replicate for the fine-tuned version, authenticating with the Replicate API, and initiating a training job using `replicate.trainings.create`. It also provides an example of how to run the fine-tuned model with a specific prompt and stop sequences.

4/21/2023

This post details the addition of new language models (StableLM, Vicuna, GPT4All) to the Replicate platform, expanding the existing 'Language Model Deployment and Fine-tuning' capability. It also mentions new playgrounds and autonomous agent projects, indicating continued innovation and user adoption in the LLM space.

4/5/2023

Introduced the capability to deploy and run large language models (LLMs) on Replicate, including official versions of FLAN-T5, GPT-J, and LLaMA. Enabled users to push custom LLMs. Launched a preview of cloud-based fine-tuning for LLMs, allowing users to train models on their own data.

3/17/2023

This post details the process of fine-tuning the LLaMA model to adopt the voice of Homer Simpson. It outlines the steps involved: acquiring a dataset of Simpsons script lines, parsing the data to create context-aware training examples (previous lines, character, and next line), modifying the training prompt to guide the model, and using a modified Alpaca codebase for fine-tuning. The post provides an example prompt and instructions for running the fine-tuned model.

3/16/2023

This post details the process of training and running Stanford Alpaca, a fine-tuned version of Meta's LLaMA model, on a user's own machine. It covers prerequisites such as obtaining LLaMA weights and having a GPU machine, cloning a modified Alpaca repository with Cog integration, converting LLaMA weights to a Hugging Face Transformers format, and executing a training script. Finally, it shows how to run the trained model using `cog predict` with an example prompt.

8/29/2022

This post details how to use the Replicate API to run the Stable Diffusion model. It covers installing the Python client, authenticating with an API token, and making predictions using the `replicate.models.get` and `version.predict` methods. It also mentions the ability to view past predictions and links to further examples and documentation.