iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
The confirmed hosted option is Mistral Large 4’s public-preview API through Mistral Studio. Mistral said on October 6, 2026, “You can try the preview API today on Mistral Studio.” The preview requires API setup, and Mistral’s quickstart says payment activation is needed to enable API keys. Availability and pricing can change; the details below reflect Mistral’s published information as of October 9, 2026.
Use the Mistral Studio preview API
Mistral’s October 6, 2026 launch announcement identifies Mistral Studio as the place to try the public-preview API. The documented route is through Mistral’s own service; the available information does not confirm another hosted provider offering Large 4 at that time. See Mistral’s launch announcement.
- Open Mistral Studio. Sign in or create an account, then locate the Mistral Large 4 public-preview API. Mistral’s announcement says the preview can be tried there.
- Set up API access if you plan to call it from an app. Follow the Mistral API quickstart. Mistral says payment activation is required to enable API keys.
- Confirm the model identifier in current documentation. The quickstart example uses the chat endpoint and the
mistral-large-latestalias, but that example alone does not confirm the correct identifier for Large 4. Check the current model documentation before putting an identifier into code: Mistral models overview. - Check your plan and usage controls. Mistral says account plans apply across Studio, API, and Vibe Code, with a shared monthly usage allowance. Whether usage can continue after that allowance depends on your pay-as-you-go settings. Review the current details in Mistral’s pricing and plan information.
What the published preview specifications and prices say
Mistral’s model page labels Large 4 as public preview and lists a 1 million-token context window. It displays the following token prices; these are vendor-published figures, not an independent cost or performance test. Check the live page and your account terms before use because both pricing and access conditions may change.
| Usage type | Displayed price |
|---|---|
| Input tokens | $0.68 per million tokens |
| Cached input tokens | $0.07 per million tokens |
| Output tokens | $2.09 per million tokens |
These figures are displayed on Mistral’s model documentation. Actual charges depend on the current rates and the plan or usage settings on your account.
#1 Best Overall
Are Large 4 weights available to run locally?
Mistral’s October 6 announcement said the weights were expected by the end of October 2026. As of October 9, that date was still in the future, so the announcement was a plan—not confirmation that weights had been released. It does not establish current local availability. Mistral’s general inference information says open-weight models can be downloaded from Hugging Face and hosted on user hardware, but that broad statement does not confirm that Large 4 weights were available there at the time. See Mistral’s inference documentation.
The launch announcement also says Large 4 would be able to run on private cloud or on premises for organizations seeking sovereign, auditable AI for security operations. It does not provide deployment requirements, timing, or purchasing terms. Mistral’s general statement that enterprise customers can deploy models on premises or in their own cloud—including AWS, Azure, and Google Cloud—does not verify Large 4 availability from those providers.
What to check before using the API
- Preview status: Confirm in Mistral Studio that Large 4 is still available to your account and remains in public preview.
- API access: Follow the current quickstart and verify whether payment activation is required for your account’s API keys.
- Model name: Use the identifier currently documented for Large 4 rather than assuming the quickstart’s
mistral-large-latestexample refers to it. - Spend controls: Check your shared allowance and pay-as-you-go settings before sending requests.
- Deployment needs: If you need private-cloud or on-premises hosting, ask Mistral for Large 4-specific availability and terms; general deployment information does not establish them.
Why published parameter counts differ
Mistral’s model documentation lists 52 billion active parameters and 1.05 trillion total parameters, as well as a 1.6-billion-parameter vision encoder. Its October 6 launch announcement instead gives 49 billion active parameters and describes the model as one trillion parameters. Those official figures conflict, and the available information does not resolve the discrepancy. They are vendor specifications, not independent performance results; there is no need to rely on a single active-parameter figure to access the preview.
Quick Recap
Rank #4
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
