Llama 4 Scout
Llama 4 Scout, a 17 billion active parameter model with 16 experts, is the best multimodal model in the world in its class and is more powerful than all previous generation Llama models, while fitting in a single H100 GPU. Additionally, Llama 4 Scout offers an industry-leading context window of 10M and delivers better results than Gemma 3, Gemini 2.0 Flash-Lite, and Mistral 3.1 across a broad range of widely reported benchmarks.
- Vision
- Tool Calling
- Structured Output
Added Sep 5, 2025
Model weightsPricing
Auto routing · per 1M tokens- Input
- $0.085
- Output
- $0.46
- Cache read
- $0.0425
Specifications
- Context window
- 328K
- Max output
- 65.5K
- Parameters
- 109B / 17B
- Total / active
Benchmarks
Benchmarks
Sourced from LMArena.
Arena Score
1321.6
Overall Rank
#264 / 421
Votes
29,775
Confidence Interval
1316.9 - 1326.3
Category Scores
Coding
#259 / 416
5,221 votes
1362.5
Math
#243 / 404
1,937 votes
1309.6
Longer Query
#265 / 399
5,794 votes
1325.4
Creative Writing
#257 / 419
3,754 votes
1288.9
Instruction Following
#271 / 421
7,468 votes
1299.3
Hard Prompts
#261 / 421
12,880 votes
1329.3
Additional Categories22
Japanese
#198 / 288
764 votes
1258.9
Polish
#204 / 243
2,467 votes
1322.1
Korean
#208 / 289
651 votes
1268.6
German
#214 / 320
800 votes
1311.1
Spanish
#228 / 304
709 votes
1316.0
French
#234 / 301
427 votes
1324.7
Industry Legal And Government
#244 / 393
1,970 votes
1341.0
Industry Mathematical
#250 / 402
1,797 votes
1314.9
Hard Prompts English
#251 / 419
5,728 votes
1352.2
Industry Medicine And Healthcare
#251 / 391
1,813 votes
1341.5
Industry Entertainment And Sports And Media
#252 / 419
5,269 votes
1288.3
Industry Writing And Literature And Language
#257 / 420
6,696 votes
1304.4
Multi Turn
#257 / 419
5,222 votes
1319.2
Russian
#257 / 385
2,266 votes
1307.8
Industry Business And Management And Financial Operations
#259 / 414
4,915 votes
1315.6
Exclude Ties
#260 / 421
21,376 votes
1268.0
English
#261 / 421
13,014 votes
1342.7
Expert
#261 / 371
1,532 votes
1308.1
Industry Life And Physical And Social Science
#264 / 419
4,964 votes
1334.4
Industry Software And It Services
#266 / 421
9,487 votes
1349.1
Non English
#268 / 421
16,750 votes
1297.6
Chinese
#270 / 395
2,077 votes
1315.1
Published 2026-11-11 · Matched as Llama 4 Scout 17B 16E Instruct · CC BY 4.0
LMArena DatasetLeaderboard results
| Benchmark | Score | Rank on NanoGPT |
|---|---|---|
Epoch AI Benchmarking Hub Epoch AI data, CC BY 4.0 · Updated Nov 13, 2026 | ||
Epoch Capabilities Index | 127.7 | #108 of 115 |
GPQA Diamond | 51.8% | #105 of 119 |
OTIS Mock AIME 2024–2025 | 7.8% | #113 of 121 |
Creative Writing v3 | 781 | #90 of 96 |
Longform Writing | 31.1 | #92 of 97 |
EQ-Bench 3 | 670 | #49 of 49 |
UGI Leaderboard UGI Leaderboard data, Apache-2.0 · Updated Oct 16, 2026 | ||
UGI Score | 31.0 | #105 of 149 |
UGI Willingness | 5.2 | #42 of 149 |
UGI Writing | 26.7 | #134 of 149 |
Filtered and ranked to models on NanoGPT.
Providers
Auto routing is available for this model. Explicit provider selection is not available.
 Recommended | $0.085 | $0.46 | $0.0425 | N/A | N/A |
Batch rows for this model use normal routing and cost the same as standard API requests. Set provider in a row to choose a provider.