Skip to main content

Model comparison

Llama 4 Maverick vs Llama 4 Scout

Data verified

Head-to-head evidence from 18 shared benchmark results across 7 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

23.49/100
Margin
16.4pts
winning →
39.87/100
0 category wins0 category wins

Public leaderboard positions: Llama 4 Maverick #191 (Supported); Llama 4 Scout #174 (Supported). Intervals and evidence labels describe ranking uncertainty, not a guarantee for a specific workload.

Evidence parity. Llama 4 Maverick and Llama 4 Scout share 18 comparable benchmark results. 0 of 8 categories are comparable. 0 results are unique to Llama 4 Maverick; 0 to Llama 4 Scout.

Updated July 20, 2026
Shared results
18
Llama 4 Maverick only
0
Llama 4 Scout only
0
Comparable categories
0 / 8

Benchmark data for Llama 4 Maverick and Llama 4 Scout is coming soon on BenchLM.

Confidence note. This is a partial-evidence comparison with 18 shared benchmark results across 7 evidence categories; 0 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

BenchLM has partial data for these models, but not enough overlapping benchmark coverage to produce a fair score-level comparison yet.

Llama 4 Scout has the larger context window at 10M, compared with 1M for Llama 4 Maverick.

Operator verdict

What I would choose

Llama 4 Scout is the better starting point when context length and serving speed matter; it lists a 10M-token window and slightly higher measured throughput. Llama 4 Maverick is the safer candidate only when its stronger result on a benchmark relevant to your workload survives a local replay.

Operator receipt. I checked 18 shared sourced rows and the operational metadata on July 15, 2026. Llama 4 Scout lists 10M of context versus 1M, with 128 tok/s versus 121 tok/s.

Honest limit. These are open-weight models, so the $0 catalog entries mean self-hosted cost varies; they do not mean inference is free. Their BenchAlign intervals are broad and overlap, so the point-score gap should not be read as a decisive capability gap.

Reviewed by Glevd on July 15, 2026.

Category breakdown

Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.

Category scores and score margins for Llama 4 Maverick and Llama 4 Scout
CategoryLlama 4 MaverickΔLlama 4 Scout
MathLlama 4 Maverick0.7MarginNo overlapLlama 4 ScoutNot measured

Decisive benchmark drivers

The largest measured benchmark gaps in this matchup, with exact reported values.

More
A · Llama 4 MaverickB · Llama 4 Scout
  1. FrontierMath v2 (Tiers 1-3)

    Math
    Source ↗
    A 0.690%B 0.000%
    Winner: Llama 4 MaverickΔ 0.7
    FrontierMath v2 (Tiers 1-3): Llama 4 Maverick scored 0.690%; Llama 4 Scout scored 0.000%. Llama 4 Maverick wins this benchmark.

Operational comparison

Runtime and commercial metrics are compared only when both models have a complete sourced value.

MetricLlama 4 MaverickLlama 4 ScoutComparison
Input / output priceUSD per 1M tokensLlama 4 Maverick$0 input / $0 outputLlama 4 Scout$0 input / $0 outputListed prices are equal.
Generation speedtokens per secondLlama 4 Maverick121 tok/sLlama 4 Scout128 tok/sLlama 4 Scout has the higher measured throughput.
First-answer latencyseconds to first tokenLlama 4 Maverick0.95 sLlama 4 Scout0.70 sLlama 4 Scout reaches the first token sooner.
Context windowmaximum listed tokensLlama 4 Maverick1MLlama 4 Scout10MLlama 4 Scout lists the larger context window.

Benchmark Deep Dive

Agentic
BenchmarkLlama 4 MaverickLlama 4 ScoutResult
AA Agentic IndexSource 1.3%1.1%Llama 4 Maverick leads
τ²-bench resultsSource 17.8%15.5%Llama 4 Maverick leads
GDPval-AASource 0.0%0.0%Tie
GDPval-AASource -1690Llama 4 Scout leads
Coding
BenchmarkLlama 4 MaverickLlama 4 ScoutResult
AA Coding IndexSource 16.3%8.2%Llama 4 Maverick leads
AA-SciCodeSource 33.1%17.0%Llama 4 Maverick leads
Reasoning
BenchmarkLlama 4 MaverickLlama 4 ScoutResult
AA-LCRSource 46.0%25.8%Llama 4 Maverick leads
CritPtSource 0.0%0.0%Tie
Knowledge
BenchmarkLlama 4 MaverickLlama 4 ScoutResult
Artificial Analysis Intelligence IndexSource 14.3%10.0%Llama 4 Maverick leads
AA-GPQA DiamondSource 67.1%58.7%Llama 4 Maverick leads
AA-HLESource 4.8%4.3%Llama 4 Maverick leads
AA-Omniscience IndexSource -41.8%-52.4%Llama 4 Maverick leads
AA-Omniscience AccuracySource 24.3%14.6%Llama 4 Maverick leads
AA-Omniscience Hallucination RateSource 87.3%78.3%Llama 4 Scout leads
Math
BenchmarkLlama 4 MaverickLlama 4 ScoutResult
FrontierMath v2 (Tiers 1-3)Source 0.690%0.000%Llama 4 Maverick leads
Multimodal
BenchmarkLlama 4 MaverickLlama 4 ScoutResult
AA-MMMU-ProSource 62.1%52.9%Llama 4 Maverick leads
Design Arena WebsiteSource 903783Llama 4 Maverick leads
Inst. Following
BenchmarkLlama 4 MaverickLlama 4 ScoutResult
AA-IFBenchSource 43.0%39.5%Llama 4 Maverick leads
Frequently Asked Questions (3)

Can I compare Llama 4 Maverick and Llama 4 Scout on BenchLM yet?

Not fully yet. BenchLM is tracking both models, but the sourced benchmark breakdown for this comparison is still coming soon.

Why does this comparison show “coming soon”?

BenchLM only shows category winners and benchmark-level calls when we have sourced results that can be compared fairly. For these models, the public benchmark coverage is not complete enough yet.

What data is available for Llama 4 Maverick and Llama 4 Scout today?

Llama 4 Maverick: $0.00 input / $0.00 output per 1M tokens Llama 4 Scout: $0.00 input / $0.00 output per 1M tokens Both model pages still include creator, context window, reasoning mode, and other metadata while benchmark coverage fills in.

Self-host vs API cost

Estimates at 50,000 req/day · 1000 tokens/req average.

Llama 4 Maverick
API / mo$0
Self-host / mo$2,610
Break-even
Llama 4 Scout
API / mo$0
Self-host / mo$2,278
Break-even
Model the full break-even

Related Comparisons

Last updated: July 20, 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.