Provider Dossier · Specialist Hardware
Groq
Category-leading tokens/sec thanks to custom LPU silicon.
The Datum Index verdict
Highly commended
Ranked No. 4 of 20 in our composite reputation index, and currently improving.
≈ 4.4 / 5 aggregate · 470 reviewer notes
Composite index · reviewed 10 July 2026
Findings
- Groq holds a Datum Index composite reputation of 87 / 100 (≈ 4.4 / 5), ranked No. 4 of 20.
- It is a specialist hardware based in Mountain View, USA, founded 2016, offering open-weight model access.
- Its flagship offering is LPU-served open models, exposed through an OpenAI-compatible API.
- Indicative pricing is $0.59 / 1M tokens in / $0.79 / 1M tokens out (medium confidence).
- Reference throughput is ~310 tok/s with ~940 ms time-to-first-token; stated uptime is 99.6%.
Groq enters our index at No. 4, a placement that reads as highly commended. The composite rests on 470 reviewer notes gathered across the public record, and it is best understood not as a score out of ten but as a standing among peers. The trend line is pointed upward: reviewer sentiment has strengthened over recent quarters.
In its favour
Category-leading tokens/sec thanks to custom LPU silicon.
On the numbers that flatter it: open-weight access, 2 named compliance attestations, and a reference throughput of ~310 tokens per second.
Points of concern
Curated, smaller model selection; capacity-gated at peaks.
The caveat a buyer should price in before committing production traffic. As with every entry, we weigh it against the field rather than against perfection.
Incident & Uptime Note
The record of being there
Groq states an availability of 99.6%. Held over a full year, that headline implies on the order of 35.0 hours of cumulative unavailability — a useful sense of scale, though real incidents cluster rather than spread evenly, and a single bad afternoon can outweigh a quiet quarter.
Observatory Panel
Latency & throughput, in distribution
Monthly panels across the models Groq serves medium
Llama 3.3 70B
| Window | TTFT ms — mean · p50 · p90 | Throughput tok/s — mean · p50 · p90 | ||||
|---|---|---|---|---|---|---|
| May 2026 | ~1170 | ~990 | ~2010 | ~282 | ~293 | ~317 |
| June 2026 | ~1030 | ~900 | ~1580 | ~310 | ~322 | ~362 |
| July 2026 | ~1090 | ~1000 | ~1690 | ~281 | ~292 | ~321 |
Llama 3.1 8B
| Window | TTFT ms — mean · p50 · p90 | Throughput tok/s — mean · p50 · p90 | ||||
|---|---|---|---|---|---|---|
| May 2026 | ~1070 | ~990 | ~1790 | ~610 | ~640 | ~680 |
| June 2026 | ~1140 | ~1020 | ~2050 | ~600 | ~620 | ~670 |
| July 2026 | ~990 | ~930 | ~1690 | ~640 | ~680 | ~760 |
All figures are single-stream, per-request rates as one user would see them — never aggregate accelerator throughput. Serving modes are not comparable with one another: shared serverless APIs batch many tenants per accelerator, dedicated and self-served deployments hand one tenant the whole card, and specialist silicon is a regime of its own. Read each figure within its mode.
No field returns on record for Groq yet — the ledger above opens as soon as the first practitioner submission clears verification.
Key Facts
The record, in brief
| Category | Specialist Hardware |
|---|---|
| Headquarters | Mountain View, USA |
| Founded | 2016 |
| Model access | Open-weight |
| Flagship / reference | LPU-served open models · Llama 3.3 70B |
| OpenAI-compatible API | Yes |
| Indicative price (in / out) | $0.59 / 1M tokens / $0.79 / 1M tokens medium |
| Reference throughput | ~310 tok/s |
| Reference TTFT | ~940 ms |
| Stated uptime | 99.6% |
| Compliance | SOC 2 Type II · GDPR |
| Composite reputation | 87 / 100 · ≈ 4.4 / 5 · 470 notes |
Reviewer Notes
What the field says
The 470 notes behind Groq's standing are drawn from the accumulated public verdict of practitioners — the recurring praises and the recurring gripes — rather than from any single survey. Two themes dominate. Admirers return to one point above all: Category-leading tokens/sec thanks to custom LPU silicon. Detractors return to another: Curated, smaller model selection; capacity-gated at peaks.
We publish neither individual reviews nor reviewer identities; the composite is an editorial synthesis, and the reviewer-note count is an order-of-magnitude indication of how much public signal informs it.
Related dossiers · see also
Peer entries in the same or adjacent category, for comparison against Groq's standing.