LOADING DATASHEET
GROUNDED IN REAL HUMAN CONVERSATIONSYOUR BENCH
Llama 4 Scout
◆ OPEN
by Meta
OPEN WEIGHTSLOCALAPILONG CONTEXTPAIDCLOSED
RANKED #209 OF 228 ASSISTANTS · OVERALL #320 OF 6,618 · VIBE SCORE 5.2 · 165 VOICES
Llama 4 Scout 17B Instruct (16E) is a mixture-of-experts (MoE) language model developed by Meta, activating 17 billion parameters out of a total of 109B. It supports native multimodal input...
HAPPY6.9
17 mentions
SAD8.1
22 mentions
VIBE TRAJECTORY
26 weeks · 82 voices
QUIETLY DEGRADING
Recent vibe and chatter are both sliding.
HOW TO USE IT
Open weights. You can use a hosted service, or download it and run it yourself, free.
RUN IT ON YOUR OWN MACHINE
Open weights: download Llama 4 Scout once and it's yours. No subscription, no rate limits, works offline.
109B parameters (MoE, 17B active), a Q4 file is about 61 GB
- BARE MINIMUM
2x NVIDIA RTX 3090 (Used)
48 GB VRAM
~$1,600 USEDQ3_K_M, ~15 tok/s, 8k context (requires tensor parallelism/split weights)
- SWEET SPOT
Mac Studio M2 Ultra (128GB Unified Memory)
128 GB UNIFIED MEMORY
~$3,000 USEDQ4_K_M or Q5_K_M, ~12 tok/s, 32k context comfortably
- FULL POWER
Mac Studio M3 Ultra (192GB Unified Memory)
192 GB UNIFIED MEMORY
~$5,500Q8_0, ~10 tok/s, full 192k context
LEARN HOW TO SET IT UP
WILL MY BOX RUN LLAMA 4 SCOUT?
The three cards above are researched picks. Search the machine you actually have — we only say yes if it should reply at a usable speed, not just load the weights and crawl.
HARDWARE PICKS & PRICES ARE RESEARCHED FROM THE LIVE WEB AND REFRESHED AUTOMATICALLY. TREAT THEM AS BALLPARK, NOT GOSPEL.
WHAT CAN YOU ACTUALLY DO WITH IT?
Everyday jobs on each of those rigs: how long Llama 4 Scout takes, and how good the crowd says it is at that kind of work.
| THE JOB | BARE MINIMUM2x NVIDIA RTX 3090 (Used) | SWEET SPOTMac Studio M2 Ultra (128GB Unified Memory) | FULL POWERMac Studio M3 Ultra (192GB Unified Memory) |
|---|---|---|---|
| Write an emaila paragraph or two, drafted from a one-line brief | ~33 S★★★★★ | ~42 S★★★★★ | ~50 S★★★★★ |
| Summarize a documenta long report boiled down to the points that matter | ~1.7 MIN★★★★★ | ~2.1 MIN★★★★★ | ~2.5 MIN★★★★★ |
| Build a websitea small landing page, markup and styles together | ~8.9 MIN★★★★★ | ~11 MIN★★★★★ | ~13 MIN★★★★★ |
| Build a backendan API with routes, storage and tests | ~28 MIN★★★★★ | ~35 MIN★★★★★ | ~42 MIN★★★★★ |
| Build a gamea playable browser game in one file | ~17 MIN★★★★★ | ~21 MIN★★★★★ | ~25 MIN★★★★★ |
STARS ARE THE COMMUNITY SCORE FOR THAT KIND OF WORK, DOCKED FOR HOW SQUASHED THE WEIGHTS GET AT EACH TIER. TIMES ARE COMPUTED FROM RESEARCHED SPEEDS. BALLPARK, NOT GOSPEL.
WHAT THE INTERNET SAYS
People are mostly frustrated with Llama 4 Scout right now.
146 POSTS · 19 COMMENTS · BLUESKY · HACKER NEWS · GITHUB · LEMMY · REDDIT · DEV FORUMS · STACK OVERFLOW · OTHER FORUMS
PEOPLE LOVE
- No fan mail yet. The crowd is still warming up.
PEOPLE COMPLAIN ABOUT
- Outages, errors, or limits get in the way (3 mentions)
TOPIC BY TOPIC
- Working when you need itA SORE SPOT
0 thumbs up · 3 thumbs down
LOOKING FOR AN ALTERNATIVE?
Checked picks first: tags say why you would switch, stars say how fully each one stands in for Llama 4 Scout.
Mistral Small
Grok 4.6> SOURCES (12)
- REDDIT · 2025-04-06 · 930 UPVOTESLlama 4 Maverick/Scout 17B launched on Lambda API
Llama 4 Maverick/Scout 17B launched on Lambda API
- REDDIT · 2026-02-17 · 587 UPVOTESCar Wash Test on 53 leading models: “I want to wash my car. The car wash is 50 meters away. Should I walk or drive?”
Car Wash Test on 53 leading models: “I want to wash my car. The car wash is 50 meters away. Should I walk or drive?”
- REDDIT · 2025-04-05 · 477 UPVOTESLlama 4 benchmarks !!
Llama 4 benchmarks !!
- REDDIT · 2025-04-05 · 279 UPVOTESLlama 4 Scout with 10M tokens
Llama 4 Scout with 10M tokens
- REDDIT · 2026-04-09 · 249 UPVOTESHow I use Cursor 10+ hours a day without torching my Claude Opus 4.6 limits
How I use Cursor 10+ hours a day without torching my Claude Opus 4.6 limits
- REDDIT · 2025-04-07 · 181 UPVOTESSeems like there was a lot of truth to this leak from 2 months ago llama 4 is beyond disappointing. it's a model that shouldn't have been…
Seems like there was a lot of truth to this leak from 2 months ago llama 4 is beyond disappointing. it's a model that shouldn't have been released.
- REDDIT · 2025-04-07 · 174 UPVOTESokay guys turn out the llama 4 benchmark is a fraud 10 million context window is fraud
okay guys turn out the llama 4 benchmark is a fraud 10 million context window is fraud
- LEMMY · 2026-05-27 · 119 UPVOTESDuckDuckGo installs are up 30% as users reject being ‘force-fed’ Google’s AI Search
DuckDuckGo installs are up 30% as users reject being ‘force-fed’ Google’s AI Search
- REDDIT · 2026-03-21 · 111 UPVOTESFree Model List (API Keys)
Free Model List (API Keys)
- REDDIT · 2025-04-06 · 101 UPVOTESIs RAG still relevant with 10M+ context length
Is RAG still relevant with 10M+ context length
- REDDIT · 2025-04-06 · 91 UPVOTESQwQ-32b outperforms Llama-4 by a lot!
QwQ-32b outperforms Llama-4 by a lot!
- REDDIT · 2025-04-07 · 85 UPVOTESLLAMA 4 Scout on Mac, 32 Tokens/sec 4-bit, 24 Tokens/sec 6-bit
LLAMA 4 Scout on Mac, 32 Tokens/sec 4-bit, 24 Tokens/sec 6-bit