Is it winning?
Mixed
Assessed capability across ten dimensions · 59.0
Loading this page
Assembling this page. Live sources are being read now.
Assembling this page. Live sources are being read now.
Wednesday, 7 October 2026
Inference neocloud built on its LPU technology; pivoted to cloud hosting after NVIDIA's ~$20B Dec-2025 licensing/acquihire transferred its core engineering team, and is raising ~$650M.
Mixed
Assessed capability across ten dimensions · 59.0
Unknown
No reputation data published
Unknown
No AI revenue disclosed
Analyst interpretationWatch-list infrastructure rather than a stable dependency; the NVIDIA acquihire leaves its roadmap unproven.
| Capability | maturityScore | status | evidenceGrade | lastVerified |
|---|---|---|---|---|
| agents | 58.3 | verified | E2 | 2026-07-20 |
| deployment_options | 51 | inferred | E1 | 2026-05-07 |
| portability | 67 | verified | E4 | 2026-07-31 |
| enterprise_assistant | 62 | verified | E2 | 2026-07-13 |
| governance | 54 | verified | E2 | 2026-07-30 |
| security | 55.9 | verified | E4 | 2026-07-31 |
| rag | 46 | inferred | E1 | 2026-05-07 |
| models | 67.6 | verified | E5 | 2026-07-31 |
| cost_controls | 67.1 | verified | E4 | 2026-07-31 |
| integrations | 61.4 | verified | E4 | 2026-07-31 |
Not covered in the capability matrix
The AI Enterprise capability matrix does not include Groq. No capability scores are shown rather than invented.
Llama 3.x hosted on Groq LPU inference, flagship open-model partnership announced at Meta Connect.
Mistral / Mixtral models hosted on GroqCloud for ultra-low-latency inference.
DeepSeek R1 / R1-distill served via GroqCloud as part of open-reasoning model lineup.
Not covered in the model inventory
The AI Enterprise model inventory records no models owned by Groq.
Not covered in the reputation seed
The AI Enterprise reputation dataset does not cover Groq.
No curated sources listed
The sourcing manifest holds no curated evidence URLs for Groq.