How many Meta models does Vector Wire track?

Meta AI models

20 MODELS TRACKED

20 published Meta models are tracked, with Muse Spark 1.3 carrying the deepest benchmark surface (66 results). Every score is source-attributed with a verification tier; every model opens its full dossier.

● LIVE ROSTER20 MODELSDATA AS OF October 7, 2026 UTCREFRESHES WITH THE PIPELINE

The rosterRANKED BY BENCHMARK SURFACE

#ModelBenchmarksStatusInput priceLast seen

01Muse Spark 1.366CURRENT$1.25/M2026-09-25

02Muse Glimmer47CURRENT$0.30/M2026-10-04

03Muse Spark 1.236CURRENT$1.25/M2026-09-22

04Muse Spark 1.1112CURRENT$1.25/M2026-09-21

05Muse Spark83UNPRICED—2026-09-21

06Llama 4 Maverick67CURRENT$0.19/M2026-09-11

07Llama 4 Scout62CURRENT$0.09/M2026-09-09

08Llama 4 Maverick 17B 128E Instruct FP810CURRENT$0.27/M2026-05-19

09Llama 3.3 70B Instruct60CURRENT$0.10/M2026-09-11

10Llama 3.2 3B Instruct59CURRENT$0.05/M2026-10-07

11Llama 3.2 Instruct 11B (Vision)22CURRENT$0.34/M2026-08-24

12Llama 3.2 Instruct 1B67CURRENT$0.03/M2026-10-07

13Llama 3.2 Instruct 90B (Vision)25CURRENT—2026-08-24

14Llama 3.1 70B Instruct70CURRENT$0.40/M2026-09-10

15Llama 3.1 8B Instruct69CURRENT$0.02/M2026-09-03

16Llama 3.1 Instruct 405B166CURRENT—2026-09-11

17Llama 3 Instruct 70B20CURRENT$0.65/M2026-09-10

18Llama 3 Instruct 8B79CURRENT$0.04/M2026-08-29

19Llama 3 (70B)27UNPRICED—2026-09-10

20Llama (65B)7UNPRICED—2026-07-02

Benchmark counts are source-attributed default-configuration results; input price is the cheapest current per-Mtok rate across tracked providers. Open a model for its full surface.

Meta head-to-headAGAINST THE FIELD'S STRONGEST

▶Muse Spark 1.17 PAIRINGS

vs GPT-5.5 vs Gemini 3.1 Pro vs GPT-5.6 Sol vs Claude Opus 4.8 vs GPT-4o vs GPT-5.4 vs Claude Opus 4.7

▶Llama 3.1 Instruct 405B4 PAIRINGS

vs Gemini 3.1 Pro vs GPT-4o vs GPT-5.4 vs GPT-5.2