Llama 4 Scout
Meta · released Apr 2, 2025 · meta-llama/Llama-4-Scout-17B-16E-Instruct
- Type
- Open weightsCustom licence
- Params
- 109B
- Context
- 1.3M
about 983K words of context · download allowed, licence restricts use
Our take
Written Aug 2, 2026Llama 4 Scout is a 109-billion-parameter text-and-image model from Meta with a 1.3-million-token request limit, released in 2025. It is built for long-document work and coding assistance, though its custom licence is more restrictive than standard open licences.
Choose this for analysing very long documents in a single pass, or for coding help where its measured skill is strongest. It is also a sensible pick when you need fast hosted inference on a budget, with several providers at the same low rate. Skip it if you need permissive licensing for redistribution or fine-tuning, if creative writing quality matters most, or if you want the cheapest possible rate and can accept slower output.
The case for it
- Extremely long request limit among downloadable models: 1,310,720 tokens, roughly ten times the 131,072 typical of many entries.
- Coding is its strongest measured skill, with a 72.9-point gap above its own creative writing score.
- Multiple cheap hosted options at identical pricing from two providers.
- Fastest throughput option is affordable: 211 tokens per second on Groq at only a modest premium over the cheapest hosts.
The case against it
- Creative writing is its weakest measured skill, 33 points below its own overall text score.
- Custom licence — not Apache 2.0 or MIT — with more restricted commercial and redistribution terms.
- Active parameter count undisclosed, so efficiency claims cannot be verified.
How good is it?
IntelligencePuzzles, maths, exam questions
Arena Text (overall)114th of 143 · 1322.9
CodingWriting and fixing code on its own
Arena Coding113th of 143 · 1362.6
Arena Coding is the only board that has scored it for this.
AgenticPlanning, calling tools, staying on task
Llama 4 Scout is not on Arena Agent (IPS), which is where the rating would come from, so there is no rating here. It is on SWE-bench Verified, in 38th of 39 with 9.1.
WritingWe do not rate this
Two boards come close and neither tests writing: Arena Creative Writing asks people which of two replies they prefer, and LiveBench Language tests whether a model understood a passage. So we show where Llama 4 Scout placed and give it no mark out of five.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done, which is why they get no rating.
Every published score for this model7 scoresEvery figure we hold, from 7 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
- Fits in memory
- weights load entirely on the card
- Spills to system RAM
- some weights offload; much slower
- Too large
- will not load even with offload
- est
- size is calculated; the verdict could change by 10%
GeForce RTX 4090 · 24 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Comfortable fit
Apple M1 Ultra (64-core GPU) · 128 GB
Room to spare. 23.9 GB spare means a 10% error in the size would not change the answer.
Memory use by level
Against a 24 GB card.
All 3 sizes here are calculated, not measured. We hold no measured file for this model, so each size comes from the parameter count and every verdict above inherits that uncertainty.
Check against your own machine → · All 71 devices, with every size →
Or rent it from someone else
Cheapest of 7 live listings. Picked at the widest standard context we hold, within one quantisation slice, so the numbers beside it are a price one host actually charges.
- per 1M tokens
- $0.10 in / $0.30 out
- Context served
- 1.3M
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| DeepInfrafp8 | $0.10 / $0.30 | 328K | not measured | Unknown | Unknown | Unknown |
| DeepInfrafp8 | $0.10 / $0.30 | 328K | 36 tok/s | No | No | Confirmed |
| OpenRouter | $0.10 / $0.30 | 1.3M | not measured | Unknown | Unknown | Unknown |
| Groq | $0.11 / $0.34 | 131K | 134 tok/s | No | No | Confirmed |
| Novita AI | $0.18 / $0.59 | 131K | not measured | Unknown | Unknown | Unknown |
| Novita AIbf16 | $0.18 / $0.59 | 131K | 8 tok/s | No | No | Confirmed |
| Google Vertex AIus-east5 | $0.25 / $0.70 | 1.3M | 53 tok/s | No | No | Confirmed |
DeepInfra
Groq
Novita AI
Google Vertex AIAcross the 7 listings we hold: 4 say they do not train on prompts, 0 say they do and 3 do not say. 4 appear in the zero-retention registry we check; the rest are unknown to us rather than confirmed either way.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
- ✓
- Supported
- ✗
- Not supported
- Not published
- host gave no parameter list
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| DeepInfrafp8 | |||
| DeepInfrafp8 | ✗ | ✓ | ✓ |
| OpenRouter | ✓ | ✓ | ✓ |
| Groq | ✓ | ✓ | ✓ |
| Novita AI | |||
| Novita AIbf16 | ✗ | ✗ | ✗ |
| Google Vertex AIus-east5 | ✓ | ✓ | ✓ |
Tool calling: 3 of 7 listings say yes, 2 say no, 2 publish no parameter list. JSON output: 4 of 7 listings say yes, 1 says no, 2 publish no parameter list. Strict schema: 4 of 7 listings say yes, 1 says no, 2 publish no parameter list.
When we formed this view
Dates behind this page
Prices last checked 14h ago
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- 2 of 7 listings publish no parameter list, so what their API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 3 of 7 listings do not say whether they train on prompts.
Licence and identifiers
What the licence allowsCustom licence, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Custom licence
This model ships custom license terms that don't map to a known template. We haven't parsed them, so commercial use, redistribution and derivatives are unverified — review the original terms before shipping.
Identifiers
- Hugging Face
- meta-llama/Llama-4-Scout-17B-16E-Instruct
- Modality record
- text+image->text
- Catalogue slug
- meta-llama-llama-4-scout