CloudLast verified June 29, 2026

Fireworks AI data retention policy

Fireworks AI runs serverless inference with zero data retention by default and does not train on your prompts or API inputs without opt-in. Summary of the verified policy plus how to route Fireworks AI through Meetily as a BYOK summary provider.

Quick policy snapshot

Default retention: See vendor policy
Zero data retention available: Yes
Trains on API customer data by default: No

Fireworks AI's API training default

Fireworks AI's privacy policy is explicit: it does not use your prompts, training data, or API inputs to train or improve its AI models without your explicit opt-in. The only training on your data is the fine-tuning you initiate yourself, and Fireworks states it does not use or share that training data for other purposes.

When Meetily users select Fireworks AI as their summary provider via BYOK, the request hits the Fireworks inference endpoint under your own API key. These defaults apply unless you have a separate enterprise arrangement that changes them.

Retention

For serverless inference on open models, Fireworks AI runs zero data retention by default. Per its documentation, Fireworks does not log or store prompt or generation data for open models without explicit user opt-in. Prompt and generation data exist only in volatile memory for the duration of the request and are not written to persistent storage. When prompt caching is active, some prompt data and the associated KV caches can sit in volatile memory for a few minutes, but this is still non-persistent. Fireworks does log metadata such as the number of tokens in a request as required to deliver the service.

There is one path to be aware of: the separate Responses API retains conversation data for 30 days when you set store=True, with an opt-out via store=False. Meetily uses the standard chat completions path, so the zero-data-retention default is what applies to transcript text routed through Meetily. For account data, Fireworks states it aims to delete your personal information within 30 days of account deletion.

Zero data retention

Zero data retention is the default for serverless inference on open models at Fireworks AI, not a contract-only add-on. That makes it a strong fit for a privacy-conscious summary path. The one way to leave that default is to deliberately opt in to logging for advanced features such as FireOptimizer; if you want the zero-retention behavior, leave those features off. Data is encrypted in transit (TLS 1.2+) and at rest (AES-256), and Fireworks documents SOC 2 Type II, HIPAA, ISO 27001, ISO 27701, and ISO 42001. Audit reports are published on the Fireworks Trust Center.

How Meetily uses Fireworks AI

Meetily routes Fireworks AI traffic through your own API key. Fireworks exposes an OpenAI-compatible endpoint at https://api.fireworks.ai/inference/v1, so you wire it into Meetily's BYOK flow by setting the OpenAI-compatible base URL to Fireworks and pasting your Fireworks API key. Meetily then sends transcript text over TLS to Fireworks for summarization, and the response is returned to Meetily and stored locally on your device. Audio is never transmitted to Fireworks at any point.

If you want the whole pipeline on-device, switch your summary model to a local provider such as Ollama. Meetily's transcription path is already local, so the local-summary route keeps both transcription and summarization on your machine.

References

"Zero Data Retention," Fireworks AI Docs (Security and Compliance), https://docs.fireworks.ai/guides/security_compliance/data_handling - accessed 2026-06-29.
"Data Security," Fireworks AI Docs (Security and Compliance), https://docs.fireworks.ai/guides/security_compliance/data_security - accessed 2026-06-29.
"Privacy Policy" (last updated November 12, 2025), Fireworks AI, https://fireworks.ai/privacy-policy - accessed 2026-06-29.
"OpenAI compatibility," Fireworks AI Docs, https://docs.fireworks.ai/tools-sdks/openai-compatibility - accessed 2026-06-29.
"Fireworks.ai Achieves SOC 2 Type II and HIPAA Compliance," Fireworks AI Blog, https://fireworks.ai/blog/fireworks-ai-achieves-soc-2-type-ii-and-hipaa-compliance - accessed 2026-06-29.
"Trust Center," Fireworks AI, https://trust.fireworks.ai/ - accessed 2026-06-29.

Last verified: June 29, 2026. Policy source: Fireworks AI policy

Frequently asked questions

Does Fireworks AI train its models on my API prompts and outputs?

No, not by default. Fireworks AI's privacy policy states it does not use your prompts, training data, or API inputs to train or improve its AI models without your explicit opt-in. Training on your data only happens if you choose to fine-tune your own model, and Fireworks states it does not use or share that training data for other purposes.

How long does Fireworks AI retain my API prompts and outputs?

For serverless inference on open models, Fireworks AI operates zero data retention by default: prompt and generation data exist only in volatile memory for the duration of the request and are not written to persistent storage. The exception is the separate Responses API, where conversation data is retained for 30 days when you pass store=True (you can opt out with store=False). Meetily uses the standard chat completions path, which is covered by the zero-data-retention default. Fireworks does log request metadata such as token counts to operate the service.

Is zero data retention available for the Fireworks AI API?

Yes, and it is the default for serverless inference on open models. Fireworks AI documents that it does not log or store prompt or generation data for open models without explicit user opt-in. If you enable certain advanced features (for example FireOptimizer), you would be opting in to logging, so leave those off if you want the zero-retention default.

Where is Fireworks AI data processed geographically?

Fireworks AI's privacy policy states it has servers for the service in the US, and that support, engineering, and other teams may support the service from the United States. For EU and UK transfers, the policy references Standard Contractual Clauses. Confirm current residency commitments in your contract and the Fireworks Trust Center before deploying a regulated workload.

Does Fireworks AI have SOC 2 or HIPAA compliance?

Yes. Fireworks AI documents SOC 2 Type II and HIPAA compliance, along with ISO 27001, ISO 27701, and ISO 42001. Audit reports and the current compliance posture are published on the Fireworks Trust Center at trust.fireworks.ai. Confirm BAA availability and scope with Fireworks directly for HIPAA-regulated workloads.

How does Meetily handle Fireworks AI when I pick it as my summary provider?

Meetily transcription is always 100% local. Fireworks AI exposes an OpenAI-compatible endpoint at https://api.fireworks.ai/inference/v1, so you select it in Meetily's BYOK flow by pointing the OpenAI-compatible base URL at Fireworks and supplying your own Fireworks API key. Meetily then sends transcript text (not audio) to Fireworks for summarization. The retention and training defaults on this page apply to your traffic, not Meetily's.

What if I want stronger guarantees than the cloud default offers?

Fireworks AI's serverless default is already zero data retention, but if you want the entire pipeline on your own machine, switch your Meetily summary model to a local provider like Ollama. Meetily's transcription path is local either way, so the local-summary route keeps both transcription and summarization on your device.

Use Fireworks AI with Meetily, on your terms

Meetily transcription stays 100% local. For summaries, bring your own Fireworks AI key (BYOK) so the data path matches the policy you just read - or pick a local model if you want zero retention by construction.

Set up BYOK Download Meetily