slopsome.com
a VRAM fit calculator and tok/s database for local LLMs
Last verified:
What is slopsome.com?
Slopsome.com is a search engine for LLM and GPU statistics that helps users compare local open-weight models and API models alongside the GPUs that can run them. The core tool answers the practical question: will model X run on GPU Y at quantization Q with a Z-token context window, and how fast will it go?
Key features include a VRAM fit-calculator that determines if a model fits in VRAM, fits with offload to CPU, works with multi-GPU setups, or won't fit at all, along with estimated tokens-per-second. The platform provides real measured throughput data (not invented numbers), side-by-side comparisons of models showing parameters, quantization sizes, minimum VRAM requirements, benchmarks, and API costs. It covers GGUF/llama.cpp/Ollama/vLLM workflows specifically.
Slopsome is built for developers, researchers, and enthusiasts running local LLMs who need to match models to their hardware. It's free to use with no signup required, sources data from real measurements rather than estimates, and includes an open read-only API for programmatic access plus a small Hugging Face Space demo.
slopsome.com pricing
Pricing model: Freemium
Free tier: Completely free to use with no signup required. No paid plans mentioned. Includes full access to VRAM fit-calculator, real tokens/sec database, side-by-side model comparisons, and all stats. Open read-only API is also free. Small Hugging Face Space demo is free to access.
slopsome.com pros
- Free to use with no signup required
- VRAM fit-calculator shows exact fit status
- Real measured throughput data, not invented numbers
- Supports multi-GPU configuration planning
- Shows offload to CPU possibilities
- Estimated tokens-per-second for each setup
- Side-by-side model comparisons
- Covers both open-weight and API models
- Shows quantization sizes for each model
- Displays minimum VRAM requirements
- Includes benchmark data for models
- Shows API cost per token for comparison
- Built specifically for GGUF/llama.cpp/Ollama/vLLM users
- Open read-only API available for developers
- Data sourced from real measurements only
- Hugging Face Space demo available
- Fast search for models and GPUs
- Shows parameters and context windows
- No invented or estimated numbers
slopsome.com cons
- No paid plans or premium features mentioned
- Read-only API has no write capabilities
- Small Hugging Face demo may be limited
- Data sourcing may miss newer models
- Focused only on GGUF/llama.cpp ecosystem
- No mobile app available
- No user accounts or saved comparisons
- Limited GPU coverage beyond common models
- No community contributions allowed
Frequently asked questions about slopsome.com
What is slopsome.com?
Slopsome.com is a search engine for LLM and GPU stats. It's a free VRAM fit-calculator plus real tokens/sec database for local open-weight and API models. You pick a model, your GPU, and quantization level, and it tells you if it fits in VRAM, with offload, with multi-GPU, or won't fit, plus estimated tokens per second.
Will model X run on GPU Y at quant Q with Z-token context?
Yes, that's the exact question slopsome.com answers. The VRAM fit-calculator tells you if a model fits in VRAM, fits with CPU offload, works with multi-GPU setup, or won't fit at all, and provides estimated tokens per second for that specific combination of model, GPU, quantization, and context size.
How fast will my model run on my GPU?
Slopsome provides real measured throughput data showing actual tokens-per-second for different model-GPU combinations. The estimated tok/s is calculated based on your specific GPU, the model size at your chosen quantization, and context window size.
Does slopsome use invented numbers or real data?
Slopsome sources data from real measurements only. There are no invented numbers. The throughput data and VRAM requirements come from actual tested measurements rather than estimates or theoretical calculations.
What quantization formats does slopsome support?
Slopsome is built for the GGUF format used by llama.cpp, Ollama, and vLLM. It shows quantization sizes for models and helps you determine which quant levels fit on your hardware.
Can I compare API models on slopsome?
Yes, slopsome includes side-by-side comparisons of both open-weight local models and API models. For API models, it shows parameters, quant sizes, minimum VRAM, benchmarks, and cost per token.
Does slopsome support multi-GPU setups?
Yes, the VRAM fit-calculator specifically checks for multi-GPU possibilities. It will tell you if a model fits with multi-GPU configuration when it won't fit on a single GPU.
Is there an API for slopsome?
Yes, there is an open read-only API available for programmatic access to the LLM and GPU stats data. This allows developers to integrate slopsome data into their own applications.
Do I need to signup for slopsome?
No, slopsome is free with no signup required. You can use all features including the VRAM fit-calculator, throughput database, and model comparisons without creating an account.
Is there a demo version of slopsome?
Yes, there is a small Hugging Face Space demo available where you can try out slopsome's features in a browser-based interactive interface before using the main site.