AI Potluck
Model components / Fine-tuned / chat models

Llama-3.1-8B-Instruct

Meta

Instruction-tuned 8B Llama 3.1 model for chat and assistant workflows, post-trained with SFT and DPO on Meta's Llama 3.1 base. The 128K context window and Llama Community License made it the default open instruct checkpoint for production chat deployments. Still the most-downloaded Llama variant on Hugging Face. Weights are gated behind a manual access request rather than an open download.

Llama 3.1 8B Instruct, released 23 Jul 2024. 8B dense, 128k context, SFT+RLHF post-trained chat model. The repo is gated `manual`, so openness was read from the model card and the license body rather than a download. Verified 2026-08-13 via the HF model card.

Openness

3 high confidence
3.0
weights
open(downloadable)
data
closed
code
partial(no training pipeline)
license
Llama-3.1-Community(non-OSI

Weights are open under Meta's Llama 3.1 Community License, which is not OSI-approved and caps commercial use at 700M monthly active users. The model is commonly marketed as open. A license that bounds commercial use rather than forbidding it caps at 3 rather than 2: the test at that boundary is whether commercial use is permitted at all, and this one permits it for everyone inside the bound, where a license that forbids commerce outright - CC-BY-NC, Mistral's Non-Production License - stays at 2. Nine other products carry a bounded license and all score 3. The repository is gated by manual approval, an access request rather than a paywall, which does not make the weights any less open; post-training is described as SFT plus DPO, with neither the pipeline nor the mixture published.

  • https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct recorded 2026-08-13

    `license:llama3.1` in the repo tags; the license body on the card reads "greater than 700 million monthly active users in the preceding calendar month, you must request a license from Meta"; repo state reads `"gated":"manual"`, an access request rather than a barrier to obtaining the weights; no training pipeline and no post-training mixture published

Adoption

4 high confidence
4.0

The one declared artifact, meta-llama/Llama-3.1-8B-Instruct, reads 7,345,657 downloads in the trailing 30 days, which bands at level 4 (1M-10M) on the software and model adoption scale. Nothing in the declared artifacts supports level 5.

Capability

3 medium confidence
3.0

A competent 2024-generation 8B chat model, a long way below the 2026 frontier and below the larger Llama 3.1 SKUs. The card's benchmark table gives Llama 3.1 8B Instruct MMLU 69.4 (5-shot, macro_avg/acc) against 83.6 for the 70B and 87.3 for the 405B, and publishes no reasoning or agentic frontier scores.

Verified 2026-08-13