Fireworks AI alternatives

Together AI

Featured

Fine-tuning · Llama / Mistral / Qwen / DeepSeek and others

8.6

Fine-tune & serve open-weight models (Llama, Mistral, DeepSeek).

Modal

Fine-tuning · Infrastructure (any model you can host)

8.7

Serverless GPUs and infra for training & serving ML.

Replicate

Fine-tuning · Thousands of community + first-party models

8.5

One-API platform for running and fine-tuning open-source models.

OpenAI Fine-tuning

Fine-tuning · GPT-4o-mini / GPT-3.5

8.4

Fine-tune GPT-4o-mini and friends on your own data.

Llama

Fine-tuning · Llama 4 (Maverick, Scout), Llama 3.3/3.2/3.1

8.3

Meta's open-weight LLM family covering 1B mobile models up to 405B frontier and natively multimodal 10M-context Llama 4 variants.

RunPod

Fine-tuning · Bring-your-own (any open-weight or custom model)

8.3

On-demand GPU cloud and serverless inference platform built specifically for AI workloads.

vLLM

Fine-tuning · Multi-model (open-weight LLMs: Llama, Qwen, DeepSeek, Mistral, Gemma, Phi, etc.)

8.3

Open-source high-throughput inference engine for serving LLMs with PagedAttention and continuous batching.

CoreWeave

Fine-tuning

8.2

AI-native GPU cloud built for large-scale training, fine-tuning, and inference on NVIDIA hardware.

Ludwig

Fine-tuning · Multi-model (PyTorch + HuggingFace Transformers)

8.2

Declarative, YAML-driven deep learning framework for fine-tuning LLMs and multi-modal models without writing training loops.

OpenPipe

Fine-tuning · Llama, Mistral, Qwen and other open-weight base models

8.2

Fine-tuning and reinforcement learning platform for turning expensive prompts into cheap, fast, task-specific models.

SGLang

Fine-tuning · Multi-model (DeepSeek, Qwen, Llama, Mistral, GLM, GPT-OSS)

8.2

Open-source high-throughput inference engine for LLMs and multimodal models with OpenAI-compatible serving.

Unsloth

Fine-tuning · Llama, Mistral, Gemma, Qwen, GLM (multi-model)

8.2

Open-source LLM fine-tuning toolkit with custom kernels that train 2-30x faster and use up to 90% less VRAM.