| title | Hosting of models for GitHub Copilot | ||
|---|---|---|---|
| shortTitle | Model hosting | ||
| allowTitleToDifferFromFilename | true | ||
| intro | Learn how different AI models are hosted for {% data variables.product.prodname_copilot %}. | ||
| versions |
|
||
| category |
|
||
| redirect_from |
|
||
| contentType | reference |
{% data variables.product.prodname_copilot %} can use a variety of AI models. This article explains how these models are hosted and served.
Used for:
- {% data variables.copilot.copilot_gpt_5_mini %}
- {% data variables.copilot.copilot_gpt_53_codex %}
- {% data variables.copilot.copilot_gpt_54 %}
- {% data variables.copilot.copilot_gpt_54_mini %}
- {% data variables.copilot.copilot_gpt_54_nano %}
- {% data variables.copilot.copilot_gpt_55 %}
- {% data variables.copilot.copilot_gpt_56_luna %}
- {% data variables.copilot.copilot_gpt_56_sol %}
- {% data variables.copilot.copilot_gpt_56_terra %}
These models are hosted by OpenAI and {% data variables.product.github %}'s Azure infrastructure.
OpenAI makes the following data commitment: We [OpenAI] do not train models on customer business data. Data processing follows OpenAI's enterprise privacy comments.
{% data variables.product.github %} maintains a zero data retention agreement with OpenAI.
All input requests and output responses processed by {% data variables.product.prodname_copilot %}'s models continue to pass through GitHub Copilot's, content filtering systems. These filters include checks for public code matches (when applied) as well as mechanisms to detect and block harmful or offensive content.
Used for:
- {% data variables.copilot.copilot_claude_haiku_45 %}
- {% data variables.copilot.copilot_claude_sonnet_45 %}
- {% data variables.copilot.copilot_claude_sonnet_46 %}
- {% data variables.copilot.copilot_claude_sonnet_5 %}
- {% data variables.copilot.copilot_claude_opus_45 %}
- {% data variables.copilot.copilot_claude_opus_46 %}
- {% data variables.copilot.copilot_claude_opus_47 %}
- {% data variables.copilot.copilot_claude_opus_48 %}
- {% data variables.copilot.copilot_claude_opus_48_fast %}
- {% data variables.copilot.copilot_claude_opus_5 %}
- {% data variables.copilot.copilot_claude_fable_5 %}
Warning
When {% data variables.copilot.copilot_claude_fable_5 %} is used, Anthropic retains data, including prompts and outputs, to operate safety classifiers that detect harmful use. Other Claude models in {% data variables.product.prodname_copilot %} remain covered by {% data variables.product.github %}'s existing data retention agreements, as documented below. Enterprise and business users need to enable the {% data variables.copilot.copilot_claude_fable_5 %} model to make it available for your organization. You can read more about Anthropic's data handling practices for this model under section F of their Service Specific Terms.
These models are hosted by Amazon Web Services, Anthropic PBC, and Google Cloud Platform. {% data variables.product.github %} has provider agreements in place to ensure data is not used for training. Additional details for each provider are included below:
- Amazon Bedrock: Amazon makes the following data commitments: Amazon Bedrock doesn't store or log your prompts and completions. Amazon Bedrock doesn't use your prompts and completions to train any AWS models and doesn't distribute them to third parties.
- Anthropic PBC: {% data variables.product.github %} maintains a zero data retention agreement with Anthropic for generally available Anthropic features in {% data variables.product.prodname_copilot %}. Some Anthropic features in beta or {% data variables.release-phases.public_preview %}—including tool search via the Messages API—are not covered by this agreement. For these features, data may be retained by Anthropic in accordance with Anthropic's ZDR documentation. {% data variables.product.github %} will update this page as ZDR coverage changes.
- Google Cloud: Google commits to not training on {% data variables.product.github %} data as part of their service terms. {% data variables.product.github %} is additionally not subject to prompt logging for abuse monitoring.
To provide better service quality and reduce latency, {% data variables.product.github %} uses prompt caching. You can read more about prompt caching on Anthropic PBC, Amazon Bedrock, and Google Cloud.
When using {% data variables.copilot.copilot_claude %}, input prompts and output completions continue to run through {% data variables.product.prodname_copilot %}'s content filters for public code matching, when applied, along with those for harmful or offensive content.
Used for:
- {% data variables.copilot.copilot_gemini_31_pro %}
- {% data variables.copilot.copilot_gemini_35_flash %}
- {% data variables.copilot.copilot_gemini_36_flash %}
- {% data variables.copilot.copilot_gemini_37_flash %}
{% data variables.product.prodname_copilot %} uses {% data variables.copilot.copilot_gemini %} models hosted on Google Cloud Platform (GCP). When using {% data variables.copilot.copilot_gemini %} models, prompts and metadata are sent to GCP, which makes the following data commitment: {% data variables.copilot.copilot_gemini %} doesn't use your prompts, or its responses, as data to train its models.
To provide better service quality and reduce latency, {% data variables.product.github %} uses prompt caching.
When using {% data variables.copilot.copilot_gemini %} models, input prompts and output completions continue to run through {% data variables.product.prodname_copilot %}'s content filters for public code matching, when applied, along with those for harmful or offensive content.
Used for:
- {% data variables.copilot.copilot_grok_45 %}
- {% data variables.copilot.copilot_grok_46 %}
These models are hosted on xAI. xAI operates these models in {% data variables.product.prodname_copilot %} under a zero data retention API policy. This means xAI commits that user content (both inputs sent to the model and outputs generated by the model):
Will not be:
- Logged for any purpose, including human review
- Saved to disk or retained in any form, including as metadata
- Accessible by xAI personnel
- Used for model training
Will only:
- Exist temporarily in RAM for the minimum time required to process and respond to each request
- Be immediately deleted from memory once the response is delivered
When using xAI, input prompts and output completions continue to run through {% data variables.product.prodname_copilot %}'s content filters for public code matching, when applied, along with those for harmful or offensive content.
For more information, see xAI's enterprise terms of service on the xAI website.
Used for:
- {% data variables.copilot.copilot_mai_code_1_flash %}
- {% data variables.copilot.copilot_mai_code_1_1_flash %}
- {% data variables.copilot.copilot_raptor_mini %}
{% data variables.copilot.copilot_mai_code_1_flash %} and {% data variables.copilot.copilot_mai_code_1_1_flash %} are first-party Microsoft models hosted on Azure in {% data variables.product.github %}'s tenant.
{% data variables.product.github %} does not use {% data variables.copilot.copilot_business_short %} or {% data variables.copilot.copilot_enterprise_short %} customer data to train AI models. For individual subscribers—{% data variables.copilot.copilot_free_short %}, {% data variables.copilot.copilot_pro_short %}, {% data variables.copilot.copilot_pro_plus_short %}, and {% data variables.copilot.copilot_max_short %} users—{% data variables.product.github %} may use {% data variables.product.prodname_copilot_short %} interaction data, including prompts (inputs), suggestions (outputs), and code snippets generated during {% data variables.product.prodname_copilot_short %} sessions to train and improve AI models, in accordance with our AUTOTITLE and applicable user settings. Individual subscribers can opt out of having their data used for AI model training. To manage this setting, see AUTOTITLE.
{% data variables.copilot.copilot_mai_code_1_flash %} and {% data variables.copilot.copilot_mai_code_1_1_flash %} are served on Microsoft Azure AI Foundry within {% data variables.product.github %}'s tenant and are subject to {% data variables.product.github %}'s data handling configuration for that deployment. For details about how data is processed, retained, and secured for models served on Azure AI Foundry, see Data, privacy, and security for Foundry Models sold by Azure in the Microsoft documentation.
When using {% data variables.copilot.copilot_mai_code_1_flash %} or {% data variables.copilot.copilot_mai_code_1_1_flash %}, input prompts and output completions continue to run through {% data variables.product.prodname_copilot %}'s content filters for public code matching, when applied, along with those for harmful or offensive content.
Open-weight models have publicly available weights. {% data reusables.copilot.open-weight-model-hosting %}
Used for:
- {% data variables.copilot.copilot_kimi_k27_code %}
- {% data variables.copilot.copilot_kimi_k3 %}
{% data variables.copilot.copilot_kimi_k27_code %} and {% data variables.copilot.copilot_kimi_k3 %} were developed by Moonshot AI. They are open-weight models that may be less aligned than other {% data variables.product.prodname_copilot_short %} models, with an elevated risk of geographic bias. Before you enable either model, review the {% data variables.copilot.copilot_kimi_k27_code %} model card or the {% data variables.copilot.copilot_kimi_k3 %} model card and run your own evaluations before enabling.
{% data variables.copilot.copilot_kimi_k27_code %} is hosted on Azure AI Foundry infrastructure managed by {% data variables.product.github %} and Microsoft.
{% data variables.copilot.copilot_kimi_k3 %} is hosted by {% data variables.product.github %} on Fireworks AI.
Both models are covered by zero data retention agreements with the hosting providers. Fireworks AI and Azure AI Foundry do not use prompts or completions to train models, and prompts and completions are not sent to Moonshot AI.
{% data variables.product.github %} does not use {% data variables.copilot.copilot_business_short %} or {% data variables.copilot.copilot_enterprise_short %} customer data to train AI models. For individual subscribers—{% data variables.copilot.copilot_free_short %}, {% data variables.copilot.copilot_pro_short %}, {% data variables.copilot.copilot_pro_plus_short %}, and {% data variables.copilot.copilot_max_short %} users—{% data variables.product.github %} may use {% data variables.product.prodname_copilot_short %} interaction data, including prompts (inputs), suggestions (outputs), and code snippets generated during {% data variables.product.prodname_copilot_short %} sessions to train and improve AI models, in accordance with our AUTOTITLE and applicable user settings. Individual subscribers can opt out of having their data used for AI model training. To manage this setting, see AUTOTITLE.
When using {% data variables.copilot.copilot_kimi_k27_code %} or {% data variables.copilot.copilot_kimi_k3 %}, input prompts and output completions continue to run through {% data variables.product.prodname_copilot %}'s content filters.
Inline suggestions, including ghost text and next edit suggestions, are powered by models hosted on Azure for {% data variables.copilot.copilot_business_short %} and {% data variables.copilot.copilot_enterprise_short %} plans. {% data variables.copilot.copilot_free_short %} and {% data variables.copilot.copilot_student_short %} user models are hosted on Fireworks AI.