I
Granite 4.1 8BOPEN
Granite 4.1 8B is a dense, decoder-only 8-billion-parameter language model from IBM, part of the Granite 4.1 family.
Prices update automatically — checked daily against provider list prices.
- $0.06 per 1M tokens, assuming 3 input for 1 output the convention used by Artificial Analysis
- cheaper than 85% of the 191 open-weight models tracked here
Benchmarks & performance
Independent benchmark scores for Granite 4.1 8B, measured by Artificial Analysis. Higher is better.
For the same budget, Granite 4.1 8B delivers 16 times more capability. Open weights also mean you can self-host it and pay nothing per token.
Models at this level cost $0.24 per 1M tokens (median of 22) — this one costs $0.06.
Cheaper and better on this index: Solar Pro 4 · Ling-3.0-flash · gpt-oss-20b and 2 more
About this model
Granite 4.1 8B is an open-weight AI model by IBM. You can download and self-host it for free; the prices below are hosted-API list prices, tracked daily, for when you prefer convenience over self-hosting.
Frequently asked questions
What is Granite 4.1 8B?
Granite 4.1 8B is an AI language model from IBM. It is open-weight: you can download it and run it on your own hardware, for free. It scores 6.4 on the Artificial Analysis intelligence index.
Is Granite 4.1 8B free?
The weights are free and open — you can self-host Granite 4.1 8B and pay nothing per token. If you prefer a hosted API, list prices are $0.05 per million input tokens and $0.1 per million output tokens.
What is Granite 4.1 8B good at?
Independent benchmarks from Artificial Analysis give it GPQA 43.3%, Humanity's Last Exam 3.8%, Long Context Reasoning 11.3%, SciCode 21.8%, IFBench 38.6%, τ²-Bench 27.8%, Terminal-Bench 3.4%, Terminal-Bench Hard 0%. It is particularly used for code generation.
How fast is Granite 4.1 8B?
It generates about 126.2 tokens per second, with a median 0.45s delay before the first token. Measured independently by Artificial Analysis.
Can I self-host Granite 4.1 8B?
Yes. Granite 4.1 8B has open weights, so you can download it and run it on your own GPU or server with tools like Ollama, vLLM or llama.cpp — with no per-token cost.