Choose your AI budget.
Know what gets billed.

GitHub's September 14 release adds three Auto preferences: efficiency, balance and intelligence. Efficiency prioritizes cost, balance weighs cost, quality and latency, and intelligence prioritizes quality. All three draw from the same available model set.

The tier is not a fixed-price package. Auto evaluates each prompt and charges for the model it selects. Even intelligence can use a smaller model for a simple task. Paid subscribers continue to receive the existing 10% discount on usage billed through Auto.

Rollout covers Visual Studio Code, Copilot CLI and the GitHub Copilot app. No fixed saving is promised for choosing efficiency: your prompts and the selected models determine the bill.

Google announced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking general availability on September 15. The current pricing table lists both under the same standard paid-tier rates.

Per million tokens, input costs $0.75 for text, $3 for audio and $1 for image/video. Output, including thinking tokens, costs $4.50 for text and $12 for audio.

Google also displays audio equivalents of $0.005 per input minute and $0.018 per output minute. They are separate input and output rates, not a single all-inclusive price per session minute. Text, image/video and optional grounding can add charges.

For Google Search grounding, the table lists 5,000 free search requests per month shared across Gemini 3.x models, then $14 per 1,000 requests. Its footnote says a model request can perform multiple search queries and each query is charged. Check query usage, not just conversation count.

This is a verified GA date paired with current published rates. We have not established an earlier comparable rate for these model IDs, so we are not calling it a price reduction.

GitHub's September 16 release adds budget-increase requests for members who exhaust their available AI credits. Previously, reaching the limit blocked features that consume credits. Now a member can request more budget when that happens.

Requests go to the organization or enterprise that pays for the budget. An organization owner, enterprise owner or billing manager can approve, adjust or deny the request. Approval sets the member's budget to the chosen amount and restores credit access.

The feature applies to Copilot Business and Enterprise under usage-based billing. It is not available for enterprises with managed users. This is an administrative spending control, not a reduction in rates or an automatic grant of free usage.