modelbenchmark.io

Meta Llama 4 Maverick 17B 128E

Meta · llama-4-maverick

Compare

Open multimodal Llama for strong reasoning with efficient everyday serving

Specification

most-agreed values

Context
131K
Max output
8K
Released
2025-04-05
Knowledge cutoff
2025-01
Retires
—
Open weights
yes
Input
text, image
Output
text

Price

US dollars per million tokens · most-agreed

Input
$0.15
Output
$0.60
Cache read
$0.05
Cache write
$0.50

Quality

6 benchmarks · 3 sources

Composite
28th
percentile of 344 scored models
Rank
249 / 344
±0.05 sd
Evidence
6 × 3
3 or more benchmarks, from 2 or more sources
Effort range
—
one configuration only
coding9th2/3 bench
reasoning60th1/4 bench
math36th3/7 bench

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
MATH level 573.0 ±1.1defaultEpoch AI2025-04-08
GPQA diamond67.0 ±2.8defaultEpoch AI2025-04-08
FrontierMath-v10.3 ±0.5default · 2 runsEpoch AI2025-04-08
OTIS Mock AIME 2024-202520.6 ±4.9defaultEpoch AI2025-04-08
Aider polyglot15.6default · aider wholeAider2025-04-06
SWE-Bench verified21.0default · mini-SWE-agentSWE-bench—

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 10 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
helicone$0.15$0.60——131K8K—
kilo · meta-llama$0.1875$0.6525$0.05—128K16K—
nano-gpt · meta-llama$0.1875$0.6525$0.05—1M66K—
openrouter · meta-llama$0.1875$0.6525$0.05—524K16K—
vercel_ai_gateway · vercel-ai-gateway$0.20$0.60——131K8K—
vercel$0.24$0.97——128K8K—
digitalocean$0.25$0.87$0.05—128K16K—
kyma · kyma$0.3344$1.337——1M16K—
databricks$0.50$1.50$0.50$0.50128K128K—
neon$0.50$1.50——1M8K—

Price history

append-only observations · a listing writes a row only when its price moves

DateHostField$/M
2026-09-02nano-gpt · meta-llamaCache read$0.075
2026-10-13nano-gpt · meta-llamaCache read$0.05
2026-09-02digitaloceanInput$0.20
2026-09-18digitaloceanInput$0.25
2026-09-02kilo · meta-llamaInput$0.20
2026-09-15kilo · meta-llamaInput$0.1875
2026-09-02nano-gpt · meta-llamaInput$0.15
2026-10-13nano-gpt · meta-llamaInput$0.1875
2026-09-02openrouter · meta-llamaInput$0.20
2026-09-15openrouter · meta-llamaInput$0.1875
2026-09-21openrouter · meta-llamaInput$0.20
2026-09-22openrouter · meta-llamaInput$0.1875
2026-09-02vercelInputfree
2026-10-22vercelInput$0.24
2026-09-02digitaloceanOutput$0.696
2026-09-18digitaloceanOutput$0.87
2026-09-02kilo · meta-llamaOutput$0.696
2026-09-15kilo · meta-llamaOutput$0.6525
2026-09-02nano-gpt · meta-llamaOutput$0.60
2026-10-13nano-gpt · meta-llamaOutput$0.6525
2026-09-02openrouter · meta-llamaOutput$0.696
2026-09-15openrouter · meta-llamaOutput$0.6525
2026-09-21openrouter · meta-llamaOutput$0.80
2026-09-22openrouter · meta-llamaOutput$0.6525
2026-09-02vercelOutputfree
2026-10-22vercelOutput$0.97
2026-09-06openrouterInput$0.20
2026-09-17openrouterInput$0.1875
2026-09-21openrouterInput$0.20
2026-09-22openrouterInput$0.1875
2026-09-06openrouterOutput$0.696
2026-09-17openrouterOutput$0.6525
2026-09-21openrouterOutput$0.80
2026-09-22openrouterOutput$0.6525

More from Meta

most-hosted first