Groq LPU Inference API Status, Latency & Model Availability (2026)

What is the latency and uptime reliability of Groq LPU Inference?

Groq LPU Inference maintains an average response latency of 65ms TTFT and an operational uptime SLA of 99.98% across its infrastructure endpoints. Supported deployment regions include ["us-east-1", "us-west-1"] with direct programmatic API access.

Verified daily via automated API latency tests and official documentation.

Zero-Shot RAG Verified

Average TTFT Latency 65ms P90 Benchmark Window

Historical Uptime SLA 99.98% 30-Day Rolling Telemetry

Regional Deployment ["us-east-1", "us-west-1"] Multi-AZ Availability

Models Available on Groq LPU Inference

GPT-5.6 Luna

99.9 IVR

Frontier foundation model developed by OpenAI featuring Sub-100ms Ultra-Efficient Edge Model architecture.

$0.18/1M in 90ms TTFT

Gemini 3.7 Flash

99.9 IVR

Frontier foundation model developed by Google featuring Native Omni Reasoning Workhorse architecture.

$0.08/1M in 75ms TTFT

Gemini 2.0 Flash

99.9 IVR

Frontier foundation model developed by Google featuring Multimodal Transformer architecture.

$0.10/1M in 380ms TTFT

DeepSeek-V4 Flash

99.9 IVR

Frontier foundation model developed by DeepSeek featuring MLA v3 Ultra-Sparse MoE architecture.

$0.14/1M in 150ms TTFT

DeepSeek-R1

99.9 IVR

Frontier foundation model developed by DeepSeek featuring MoE Reasoning architecture.

$0.55/1M in 1800ms TTFT

DeepSeek-V3

99.9 IVR

Frontier foundation model developed by DeepSeek featuring Multi-head Latent Attention MoE (671B/37B) architecture.

$0.14/1M in 340ms TTFT

Qwen 3.8 Flash Next

99.9 IVR

Frontier foundation model developed by Alibaba Cloud featuring High-Throughput Asian/Global MoE architecture.

$0.12/1M in 120ms TTFT

GLM 5.3 Flash

99.9 IVR

Frontier foundation model developed by Zhipu AI featuring Agent-Centric Fast MoE architecture.

$0.15/1M in 130ms TTFT

Codestral 25.01

99.9 IVR

Frontier foundation model developed by Mistral AI featuring Dense 256k Code Specialist architecture.

$0.30/1M in 150ms TTFT

Microsoft Phi-4 (14B)

99.9 IVR

Frontier foundation model developed by Microsoft featuring Dense Small Language Model architecture.

$0.12/1M in 110ms TTFT

Llama 3.3 70B Instruct

99.9 IVR

Frontier foundation model developed by Meta featuring Dense Transformer architecture.

$0.18/1M in 420ms TTFT

Qwen 2.5 72B Instruct

99.9 IVR

Frontier foundation model developed by Alibaba Cloud featuring Dense Transformer architecture.

$0.35/1M in 420ms TTFT

Qwen 2.5 Max

99.9 IVR

Frontier foundation model developed by Alibaba Cloud featuring MoE architecture.

$0.28/1M in 480ms TTFT

o3-mini

77 IVR

Frontier foundation model developed by OpenAI featuring Reasoning MoE architecture.

$1.10/1M in 1200ms TTFT

xAI Grok 4.6

74.6 IVR

Frontier foundation model developed by xAI featuring Colossus 200k-Cluster Real-Time Frontier architecture.

$2.00/1M in 280ms TTFT

GPT-5.6 Terra

72.5 IVR

Frontier foundation model developed by OpenAI featuring Balanced Production Workhorse MoE architecture.

$1.50/1M in 210ms TTFT

Claude 3.5 Haiku

70.4 IVR

Frontier foundation model developed by Anthropic featuring Lightweight Dense architecture.

$0.80/1M in 140ms TTFT

Composer 2.5

70.4 IVR

Frontier foundation model developed by Anysphere / Cursor featuring Full-Repository Diff & Refactor Specialist architecture.

$1.80/1M in 180ms TTFT

Fable 5

59.8 IVR

Frontier foundation model developed by Fable Studio featuring Generative Simulation & World Agent Engine architecture.

$2.00/1M in 240ms TTFT

Mistral Large 2

53.9 IVR

Frontier foundation model developed by Mistral AI featuring Dense Transformer architecture.

$2.00/1M in 550ms TTFT

Claude 3.7 Sonnet

49.5 IVR

Frontier foundation model developed by Anthropic featuring Hybrid Reasoning architecture.

$3.00/1M in 650ms TTFT

Claude 3.5 Sonnet

46 IVR

Frontier foundation model developed by Anthropic featuring Dense Multimodal Transformer architecture.

$3.00/1M in 320ms TTFT

Grok 3

45.9 IVR

Frontier foundation model developed by xAI featuring Dense Transformer architecture.

$3.00/1M in 850ms TTFT

OpenAI GPT-4o

45.5 IVR

Frontier foundation model developed by OpenAI featuring Omni Multimodal Transformer architecture.

$2.50/1M in 280ms TTFT

Claude Opus 5

42.7 IVR

Frontier foundation model developed by Anthropic featuring Next-Gen Ultra-Dense Hybrid Reasoning architecture.

$5.00/1M in 340ms TTFT

GPT-5.6 Sol

35.3 IVR

Frontier foundation model developed by OpenAI featuring Flagship Autonomous Reasoning Cluster architecture.

$8.00/1M in 420ms TTFT

OpenAI o1

21 IVR

Frontier foundation model developed by OpenAI featuring Large-Scale Reasoning Model architecture.

$15.00/1M in 850ms TTFT