The Latent
EN▾
EnglishEspañol中文PortuguêsFrançaisالعربية日本語한국어
Sign Up
NEWSDATARESEARCHPODCASTNEWSLETTERPRESSPARTNER WITH US →
Data/Capability

Capability

Historical Inference Speeds

Historical Inference Speeds. Demo series compiled by The Latent research desk.

Historical Inference Speeds

  • Groq
  • Google
  • OpenAI
  • Anthropic
Historical Inference SpeedsDemo preview through July 2026 (tokens per second).3K2K1K0Aug '25Oct '25Dec '25Feb '26Apr '26Jun '26Groq: 2.3K on Jul '26. Google: 841.6 on Jul '26. OpenAI: 621.2 on Jul '26. Anthropic: 561.1 on Jul '26
SOURCE: The LatentUPDATED: Jul 1, 2026
ZOOMALLYTD12M3M1M
CSV
View data
DateGroqGoogleOpenAIAnthropic
2025-08-011899.53694.38512.99463.77
2025-09-011796.83656.37484.56437.76
2025-10-011885.67688.65508.26459.05
2025-11-011862.12680.19502.13453.61
2025-12-011979.8722.35532.64480.63
2026-01-012181.21796.5587.81530.84
2026-02-012129.91778.16574.57519.16
2026-03-012182.04796.88588.15531.2
2026-04-012156.32788.2582.27526.37
2026-05-012036.2743.7548.96495.88
2026-06-012218.42809.66597.19539.03
2026-07-012304.47841.64621.22561.11

API access · Pro coming soon

Demo preview through July 2026 (tokens per second).

Download CSVPro API coming soon

Methodology

Synthetic demo series generated to preview the data product layout.

Related charts

  • Historical Token Throughput per GPU
  • Historical Token Efficiency Across Frontier Models
  • Epoch Capabilities Index (ECI)

The Latent

AI industry news. A sister publication to The Block.

Editorial

  • Standards
  • Corrections
  • Commercial policy
  • Contact

Company

  • About
  • Privacy Policy
  • Terms of Service
  • Security
  • The Block