DEPLOYDatabase

AI chip · Groq

Groq LPU (Language Processing Unit)

Deterministic-compilation inference accelerator optimized for low-latency LLM token generation; SRAM-based memory architecture.

Market position

The inference speed record-holder: Llama 70B at 500+ tokens/sec per user through 2024, an order of magnitude ahead of anyone else, by moving all weights into SRAM. Zero HBM, deterministic latency. Weakness: no training, and per-chip capacity is small so real deployments need dozens of racks.

DEPLOY editorial. What the vendor PDF cannot tell you.

Physical-AI cross-link

Derived from claims already on record (vendor first-party TDP, throughput, price, and the reference-design cluster registry). The arithmetic is shown per row so it can be audited. No estimated inputs.

Perf per watt
0.50 FP16 TFLOPS/W
188 TFLOPS ÷ 375 W = 0.50

What fits in 0.23 GB

Weights-only footprint for public open-weight LLMs. Pure arithmetic: params × bytes/param. Excludes KV cache and activation memory; add ~10-30% headroom for real serving. A model that does not fit at FP16 may still fit at INT8 or INT4 with quality trade-offs. Not a benchmark.

ModelParamsFP16INT8INT4
Llama 3.1 8B
dense
8 B16 GB ✗8 GB ✗4 GB ✗
Llama 3.1 70B
dense
70 B140 GB ✗70 GB ✗35 GB ✗
Llama 3.1 405B
dense
405 B810 GB ✗405 GB ✗203 GB ✗
Llama 3.3 70B
dense
70 B140 GB ✗70 GB ✗35 GB ✗
DeepSeek V3
MoE (671B total, 37B active per token)
671 B1342 GB ✗671 GB ✗336 GB ✗
DeepSeek R1
MoE (671B total, 37B active per token)
671 B1342 GB ✗671 GB ✗336 GB ✗
Qwen 2.5 7B
dense
7 B14 GB ✗7 GB ✗4 GB ✗
Qwen 2.5 72B
dense
72 B144 GB ✗72 GB ✗36 GB ✗
Mixtral 8x7B
MoE (46.7B total, 12.9B active per token)
46.7 B93 GB ✗47 GB ✗23 GB ✗
Mixtral 8x22B
MoE (141B total, 39B active per token)
141 B282 GB ✗141 GB ✗71 GB ✗
Gemma 2 27B
dense
27 B54 GB ✗27 GB ✗14 GB ✗
Command R+
dense
104 B208 GB ✗104 GB ✗52 GB ✗
Kimi K2
MoE (1T total, 32B active per token)
1000 B2000 GB ✗1000 GB ✗500 GB ✗

Math: FP16 = params × 2 bytes; INT8 = params × 1 byte; INT4 = params × 0.5 bytes. MoE models sum every expert (full weights on disk), not the per-token active subset.

Common questions

Answer-first, sourced. Every claim below traces to a specific field on this page or a cited datasheet. FAQPage schema is emitted so LLM crawlers can lift these verbatim.

How much does Groq LPU (Language Processing Unit) cost?

Groq LPU (Language Processing Unit) launch pricing is not on record in the DEPLOY registry. Vendor list prices are typically published in the vendor's own briefing rather than the datasheet.

How much memory does Groq LPU (Language Processing Unit) have?

0.23 GB of SRAM (230 MB on-die) at 80,000 GB/s. Vendor datasheet, first-party.

How much power does one Groq LPU (Language Processing Unit) draw?

375 W TDP (thermal design power) per chip. System-level draw is higher: see the reference-design section for rack-level kW.

Which open-weight LLMs fit on one Groq LPU (Language Processing Unit)?

Of the 13 public open-weight LLMs DEPLOY tracks: 0 fit at FP16, 0 at INT8, 0 at INT4 (weights only, excludes KV cache). The full table is above with per-model math. Sparsity, offloading and multi-GPU serving change the picture; this row is single-chip weights-only.

Key facts

Class
Compute SoC (AI accelerator)
Designer
Groq
Safety-critical?
No (data-center inference / training)
Record as of
2026-08-25
Specifications (9 fields, click to expand)

Vendor datasheet figures (first-party). Dense throughput first; sparse (2:4) in parentheses where NVIDIA quotes it. The exhaustive spec sheet lives on the datasheet URL below the table: this row set covers what buyers actually compare on.

Process node
GlobalFoundries 14nm
TDP
375 W
Memory
0.23 GB SRAM (230 MB on-die)
Memory bandwidth
80,000 GB/s
FP16 (dense)
188 TFLOPS
INT8 (dense)
750 TOPS
Form factor
PCIe (GroqCard) / rack-scale GroqRack
Announced
2020-01-01
Released
2021-06-01

Source: vendor datasheet

Compare with

See every chip comparison →

Sources

Adoption rows appear as we verify chip-integration claims per model. Every row is a public claim the maker or a tier-1 source has stated; verification tier (verified / reported / inferred) is shown per row. See every chip on record for the full catalog or the Groq record.