Llama 4 Scout 17B 16E Instruct: documented context capacity is 10 million tokens
Meta pushes the boundaries of information processing with a massive 10 million token context capacity.
Documented context capacity is 10 million tokens — verified by the OnAirToday AI Progress Tracker against Meta AI's published documentation.
OnAirToday
Newsroom
The OnAirToday AI Progress Tracker recorded a verified change to Meta AI's Llama 4 Scout 17B 16E Instruct (version Llama-4-Scout-17B-16E-Instruct): Documented context capacity is 10 million tokens.
Meta's Llama 4 Scout is an instruction-tuned, multimodal mixture-of-experts model with 17 billion active parameters, 16 experts and downloadable weights. Its published specifications include a 10-million-token context window and text-and-image inputs, while hosted inference pricing depends on the provider.
What changed in this review:
- Added Meta-reported MMLU-Pro, GPQA Diamond and MMMU results; these are distinct from independently reported evaluations.
- Documented multimodal capabilities, deployment requirements, licensing limitations and safety risks.
- Capability score 33 → 66
- Flaws and limitations updated
- Risks and threats updated
Teams building on Llama 4 Scout 17B 16E Instruct should confirm limits, pricing and benchmark coverage against the vendor's own documentation before relying on these figures.
The full profile — capabilities, benchmark citations, known flaws and risk guidance — is on the tracker at /ai-models/llama-4-scout-17b-16e-instruct.
References:
- https://huggingface.co/meta-llama/Llama-4-Scout-17B-16E-Instruct
- https://ai.meta.com/blog/llama-4-multimodal-intelligence/
Generated insight by OnAirToday AI
The jump to a 10 million token context window signifies a transformative shift in how LLMs process vast, multi-document datasets for long-term reasoning. This capability effectively bridges the gap between simple chatbots and autonomous systems capable of analyzing entire enterprise knowledge bases in a single prompt.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.