PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteiTechGuides 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
Short answer: Use AI.COUNT_TOKENS before a run to estimate input tokens, then open the completed query’s Query results → Job information pane to record actual input, output, thought and cached-token counts. Match Vertex AI charges in Cloud Billing with the bigquery_job_id_prefix label, wait for the possible 24-hour posting delay, and add ordinary BigQuery bytes or slot charges to the model-token estimate. Treat daily quotas as execution guardrails—not as a monthly bill—and monitor embedding functions separately because they use a different quota and billing path.
What you are measuring
BigQuery AI functions can incur two distinct categories of cost. The query itself can consume BigQuery bytes processed or slots, while a remote model endpoint can charge for inference. A useful forecast therefore has to join observed token usage with the model’s current rates and with the BigQuery cost of the statement that called the function. Looking only at a token counter or only at BigQuery bytes gives an incomplete answer.
BigQuery offers general-purpose functions such as AI.GENERATE, AI.GENERATE_TEXT and AI.GENERATE_EMBEDDING. It also offers managed functions including AI.IF, AI.SCORE, AI.CLASSIFY and AI.AGG. These functions do not all use the same service. General Gemini inference is sent to Gemini Enterprise Agent Platform; embeddings and task-specific functions can follow different services and billing structures.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Generation and embedding are not interchangeable
| Workload | Examples | What to inspect | Quota and billing note |
|---|---|---|---|
| Generative inference | AI.GENERATE, AI.GENERATE_TEXT, managed text functions |
Job information input, output, thought and cache token counters | Gemini-backed generation uses the generative token quotas; model pricing determines which token categories are billable |
| Embeddings | AI.EMBED, AI.GENERATE_EMBEDDING |
The function’s returned statistics, including token_count when provided |
Excluded from the generative-inference token quotas and subject to different billing structures |
| Managed or task-specific functions | AI.IF, AI.SCORE, AI.CLASSIFY, AI.AGG |
The function’s documented service, model and the resulting job and billing records | Do not assume that a generation quota or rate applies without checking the function’s current documentation |
The generative-AI overview published by Google Cloud and updated October 13, 2026, defines the input counter as the total input tokens sent by all generative AI functions called in the query. That wording matters: the number is query-wide, not a count from one selected row or one function invocation.
#1 Best Overall
- Powerful Turbo Fan:CAGIWIRU Electric Air Duster have reached speed up to 130000RPM,Can efficient remove dust and debris.With LED light and Three Adjustable Speed to meet with different cleaning tasks.
- With Attached USB C Fasting Charging cable.Charing 3 hours,Enjoy up to 240 minutes of use on the lowest setting, For third speed can use for 25mins.with four charging options to meet with your needs.
- Wide Application: attached 5 different nozzles,making it suitable for a variety of scenes, such as car,pet, pc, keyboards, and other electronic devices. It also serves for office and home clean duster.
- Handy Design & Portable:Cordless air duster electric nozzle is Light weight and small size, design and comfortable to hold and space saving, convenient to carry around. You can put it in your room or inside the car without taking up space.
- Economical &Reusable: Built durable materials with a long-lasting battery,our CAGIWIRU air duster it's a sustainable alternative to disposable air cans.
Preflight: estimate input tokens before execution
Run AI.COUNT_TOKENS against the prompt or a representative input before launching a large query. This is a planning measurement, not a final invoice. It tells you how much input text the supplied sample would tokenize, allowing you to catch an unexpectedly large template, document or multimodal payload before the generation starts.
Step 1: choose a representative sample
Use inputs that reflect production. Include the same system instructions, user prompt template, columns concatenated into the prompt, language mix and media types that the real query will send. A short test string can understate usage when production rows contain long documents or repeated context.
Step 2: run the token-counting expression
Use AI.COUNT_TOKENS in a small query or a controlled sample. Keep the model and input shape aligned with the function you plan to call. Store the result with the sample size and prompt-template version so that later changes can be explained.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Step 3: scale only after checking row behavior
Do not multiply one token estimate by the total row count until you have checked the distribution of row lengths. A few unusually large rows can dominate a monthly total. Record at least a median and a high percentile of input tokens by workload; those observations become the low and high cases in your forecast.
AI.COUNT_TOKENS does not predict every token that will appear in a completed job. Actual totals can change with the number of rows, prompt-template expansion, multimodal inputs, requested output length, candidate count and model reasoning settings. Output and thought tokens must be measured from representative completed jobs.
Capture authoritative usage from a completed job
After the query finishes, open the BigQuery console, select the query result, and choose Job information. For Gemini models other than embedding models, record all four counters shown for the job:
Rank #2
- Screen Cleaner is Your Screen’s New Bestfriend! - EVEO is proud to present a screen cleaner that’s perfectly suited for your TV. Meet your new screen cleaner, computer screen cleaner, laptop cleaner, iPad cleaner, Macbook,tv cleaner or any other electronic device. Our screen cleaner spray and wipe along with the included plush microfiber cloth, easily removes dust, fingerprints & other smudges on the screen’s surface. Clean effectively for a streak-free screen
- Gentle on your Screens - Our screen cleaner spray and wipes is compatible to be used even for the most sensitive LCD, LED, CRT, and OLED screens in proper use. Be at ease knowing that this product is compatible with all the major brands: LG, SONY, Samsung, Sharp, iPad, iPhone, Android screens, Kindle, PC monitors - we’ve got you covered! Can be used as LCD screen cleaner, tv cleaner, smart tv screen cleaner, and more
- Premium Super Soft Plush Microfiber Cloth - Included in this efficient screen cleaning kit. This screen cleaner spray and wipe has our signature plush microfiber cloth specially designed to comply with screens and monitors and leave them streak-free. Has been shown to effectively absorbs stubborn stains, zero streaks, and no fiber shedding. Always use this together with our screen cleaner for the best results
- Easy-to-Use Screen Cleaner Kit + microfiber cleaning cloth - The EVEO Screen Cleaning Kit will easily be your new favorite tool when it comes to taking care of the things you love most, quick and easy solution for a streak-free screen. Simply screw on the sprayer gun on the screen cleaner bottle. Spray the cleaning liquid onto the microfiber cloth, then wipe the screen until desired cleanliness & shine is achieved.
- Quality Screen Care Kit Worth Purchasing - Enjoy a convenient Screen Cleaner anytime, anywhere you need it. Your satisfaction is our top priority. We stand behind the quality of our products and that’s why, you shouldn’t be worrying at all. We manufacture the Best Screen Cleaner. and can be used for laptop cleaning kit, laptop screen cleaner, tv screen cleaner for smart tv. MacBook screen cleaner, monitor cleaner or screen cleaner spray for electronic devices with microfiber cleaning cloth
- Input token count: the total input tokens sent by all generative AI functions in that query.
- Output token count: tokens across candidate responses.
- Thought token count: model-generated reasoning tokens when the model exposes them.
- Cache token count: implicitly cached input tokens.
These are the observed totals to use for reconciliation. Preserve the query job ID, execution date, model, region, prompt-template version, row count and whether retries occurred alongside the counters. Without that context, a later cost difference can look like a pricing problem when it was actually a workload change.
Why output and thought tokens can surprise you
Input is often the easiest quantity to estimate because you control the prompt. Output varies with the response length and with the number of candidates requested. Thought tokens can add usage when the selected model performs internal reasoning. A prompt that appears unchanged can therefore produce a higher billed total after a model or reasoning-setting change. Cache tokens identify input that was implicitly reused; whether they are billed, and at what rate, is model-specific.
Attribute model charges to the BigQuery job
Use Cloud Billing to connect endpoint charges with the query that generated them. Filter the service to Vertex AI, then filter by the bigquery_job_id_prefix label. Google documents a prefix rule: the value can be at most 63 characters, and it must be lowercase when the original job ID contains uppercase characters. Use the documented prefix for the job rather than inventing a new label value.
Some charges can take up to 24 hours to appear in Cloud Billing. A same-day comparison is therefore provisional. Keep the job’s observed counters and the billing export or console result together, then reconcile again after the posting window. The Google Cloud generative-AI overview explicitly warns that some charges may take up to 24 hours to appear.
Separate endpoint and BigQuery charges
Remote-model inference does not replace the normal BigQuery charge. The statement can still incur bytes-processed or slot costs while the model endpoint records token-based charges. In your reconciliation, maintain separate columns for:
- Vertex AI or other endpoint token charges.
- BigQuery on-demand bytes processed, if that pricing model applies.
- BigQuery slot consumption or reservations, if the project uses slot-based pricing.
- Any additional API or endpoint charge documented for the selected function or service.
Do not interpret a Vertex AI line item as the entire cost of the query, and do not attribute ordinary BigQuery scan cost to model tokens.
Rank #3
- 【Large size】：Measuring 800 x 300 mm/ 31. 5×12 inches, which is wider and taller than others. Large mouse pad can perfectly fit your keyboard and mouse of any size, and there is also ample space for peripherals such as phones, tablets, etc.
- 【LED backlight mouse pad】：Blade Hawks RGB mouse pad provides 7 static modes and 7 dynamic modes will give you the options you are looking for. Power-off memory function remembers the last lighting mode you selected. The super glow fiber Chroma will give you a colorful gaming setup that you want.
- 【Anti-slip rubber base】：Made from Natural rubber, this gaming mousepad will firmly grip your desktop, allowing you to enjoy the ultimate precision in the games you are most passionate about. It also comes with a texture that enhances this ability even further.
- 【Smooth and waterproof surface】: Micro-textured cloth surface, for extremely precise mouse control and a smooth movement. When liquid splashes on the gaming mouse pad, it forms water droplets and slides down. Will not delay your work or gameplay.
- 【Plug and play】: This mouse and keyboard pad is powered by USB, plug and play, no driver required. One button control light modes. Press one time to change the lighting mode, press twice to change the brightness, press and hold about 3 seconds turn ON/OFF.
Build a monthly forecast that can be audited
Create one row per model and workload, such as summarization, classification or extraction. Keep generation and embedding rows separate. At minimum, record the expected monthly calls or rows, input tokens, output tokens, thought tokens, cache tokens, the model’s published rates, and the BigQuery bytes or slot usage associated with the statement.
| Forecast field | How to obtain it | Why it matters |
|---|---|---|
| Workload and function | Query text and function name | Different functions can use different services and billing paths |
| Model and endpoint | Job configuration and model documentation | Rates, regions and quota behavior are model-dependent |
| Monthly rows or calls | Scheduled-query history, pipeline volume or an explicit growth assumption | Converts per-call observations into a month total |
| Input, output, thought and cache tokens | Job information counters; use AI.COUNT_TOKENS only for preflight input estimates |
Separates the token categories that pricing may treat differently |
| Token rates | The current model pricing page | Pricing is dynamic and not universal across AI functions |
| BigQuery bytes or slots | Job details, query execution information and the project’s pricing model | Captures the non-endpoint part of the query cost |
| Posting and reconciliation status | Cloud Billing after the possible 24-hour delay | Prevents premature conclusions about a mismatch |
Token-cost equation
Use this budgeting equation with the selected model’s published rate units:
monthly model cost = (input_tokens × input_rate + output_tokens × output_rate + thought_tokens × thought_rate + cache_tokens × cache_rate) / 1,000,000
Free tools Windows power users keep installed
One-click scans. No signup required.
Include only token categories that the model’s current pricing page actually bills. Some models can publish different rates for input, output, reasoning or cached input, while another model may omit one of those categories. If a rate is not listed for a category, do not silently substitute a different rate; mark that category as not billed or not stated by the applicable pricing documentation and verify before budgeting.
Then add the estimated BigQuery bytes or slot charge and any other endpoint or API charge. Google does not publish one universal monthly price for all BigQuery AI functions, so a forecast must be assembled from the chosen function, model, region and workload rather than copied from a single calculator.
Use three traffic scenarios
| Scenario | Token and volume basis | Use |
|---|---|---|
| Low | Observed median rows and token counts | A conservative floor for ordinary traffic |
| Expected | Rolling 30-day average by workload | The operating budget and reconciliation target |
| High | p95 rows and token counts plus an explicit growth assumption | Capacity and spend protection during spikes |
Recalculate the scenarios when the prompt template, model, candidate count, reasoning setting, region or row mix changes. Reconcile the expected scenario to Cloud Billing after charges post. If the billed amount differs, inspect the model and region first, then cache behavior, retries, row-level failures, and BigQuery bytes or slots.
Rank #4
- 【Softer and More Comfortable】Vaydeer wrist rest has unique diamond pattern, which is the combination of softness and aesthetics. The materials of wrist rest are improved into higher quality memory foam and covered with silky smooth lycra. The computer wrist rest makes you as comfortable and cushiony as like rest your wrists on clouds.
- 【Ergonomic Wrist Saver】The wrist rests for keyboard and mouse comes with a 17.32×3.15×0.83 inch keyboard wrist pad and a 5.94×3.15×0.83 inch mouse wrist support. Based on ergonomic design, the unique concave shape is the perfect fit for your wrist joints. The wrist rest pad fits most computer keyboards and laptops, improve hand and wrist posture, release your wrist and arm stress.
- 【Non-Slip Rubber Bottom】Featuring an anti-skid silicone base on the bottom, this wrist keyboard support stays firmly in place on your desk, preventing the padding from sliding around, ensuring stable and consistent wrist support during extended computer sessions.
- 【Better Experience & Pain Relief】Our keyboard arm rest is beneficial to alleviate the soreness caused by direct contact and friction between your arm and a hard desk surface, reducing the risk of wrist fatigue or carpal tunnel. The soft texture of memory foam can evenly distribute the pressure around your wrists and provide good support with just enough give.
- 【Helpful in Multiple Scenarios】Whether you're working, studying, writing, typing, gaming, this keyboard and mouse rest combo is an essential accessory to add comfort and support to your hands and wrists. It’s also a great gift for men, women, family, friend, coworker, gamer, teacher, etc.
Why a BigQuery AI query costs more than expected
Prompt growth
Joining additional columns, appending conversation history or embedding a full document instead of a short excerpt can raise input tokens per row. Compare the current prompt template with the version used when the forecast was made, and run AI.COUNT_TOKENS on representative long rows.
Recommended Free Tools
More output or candidates
A higher output limit or multiple candidate responses increases output-token totals. Check the completed job’s output counter rather than assuming that the prompt alone determines the bill.
Reasoning tokens
Models that expose thought tokens can consume additional billable tokens. A model or reasoning-setting change can raise thought counts without changing the visible prompt. Compare job information across the two configurations.
Retries and partial failures
Retries can create additional model calls. Row-level failures can also make a run’s successful output look smaller than its billed activity. Review query logs, request-response logs if enabled, and the job status when totals do not match the expected row count.
Cache assumptions
Cached-token counts are reported separately. Do not assume that every cached token is free or that the same cache treatment applies to every model; use the model’s pricing terms. Cached tokens do not count toward the documented generative token quotas, but that quota exemption is not a statement that all cached usage is unbilled.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Billing-posting delay
Wait up to 24 hours before treating a missing Vertex AI line item as zero cost. Compare like-for-like windows after the delay has elapsed.
Best Value
- 【No Drill Mounted】Cable management tray can be assembled in as little as 3 minutes or less-simply. Side clips easily clampe the edge of desk, pass through the holes on the sides, and secure excess cables with baskets. Desk cable management supports inward or outward installation to meet your needs in different scenarios, and provides you with great convenience when collecting and organizing wires.
- 【Rubber Pad Protects Desk】Wire organizer under desk built-in soft rubber pads on the clamp protect your desk from scratches, so you will not damage your furniture or office.
- 【Sturdy and Beautiful】Cord management under desk is made of high-quality carbon steel and holds about 15 lbs. The wire basket has 2 holes on each side of the tray to enable your cables to pass freely. Cable baskets are suitable for desk thickness use from 0.4" to 2.4".
- 【Space Saving & Keep Organized】Desk cable management tray holds all your power cords, outlet strips, USB hubs and computer transformers hidden from your feet, and the data cables will not enter the vacuum cleaner to keep your desk clean and tidy. Desk wire organizer is also suitable for use in the kitchen and outdoors.
- 【Reduce Safety Risks】Cable tray under desk keeps all plugs away from your kids, no more worry about tripping. Using 2 holes wire tray saves extra space, helps you easily cabling and wire protection. The mesh design is not easy to accumulate dust, but avoids safety accidents caused by messy wires.
BigQuery work outside the model call
A query can scan a large table, sort or join data, and then send only a subset of rows to the model. The endpoint token count may be reasonable while BigQuery bytes or slot usage drives the unexpected total. Inspect both billing paths.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Set token quotas as guardrails
Google Cloud documents these default daily quotas for Gemini-backed generative AI functions in 2026:
| Quota | Default daily amount | Scope |
|---|---|---|
| AI input tokens per day | 200,000,000,000 | Google Cloud project, global across regions |
| AI input tokens per day per user | 150,000,000,000 | Individual user, global across regions |
| AI output tokens per day | 20,000,000,000 | Google Cloud project, global across regions |
| AI output tokens per day per user | 15,000,000,000 | Individual user, global across regions |
These are daily service limits, not a monthly invoice forecast. They are tracked globally across regions and reset at midnight Pacific time. A workload can stay below a daily quota and still exceed its monthly budget, or hit a daily quota while its month-to-date spend is modest.
Impose a stricter project limit
- Open IAM & Admin → Quotas & System Limits in Google Cloud.
- Filter the quota list for the BigQuery API.
- Locate the AI input or output token quota that applies to the project or user scope.
- Create a quota override with the lower limit your operations team can safely support.
- Document the owner, reset time, affected regions and the reason for the limit, then test a controlled query before relying on it as a production cap.
Quota enforcement occurs before execution and during execution. When a configured quota is exhausted, new model calls can fail with QuotaExceeded; remaining rows in an in-progress query can fail, and subsequent generative-AI queries can remain blocked until the midnight Pacific reset. Build retry and alerting logic around that behavior instead of assuming that a query will simply stop after a clean row boundary.
Provisioned-throughput models are not billed by token usage. If you use that billing mode, quotas still protect execution capacity, but a token-based spend cap is not a substitute for reviewing the provisioned-throughput commitment and its utilization.
Embedding functions require a separate monitor
AI.EMBED and AI.GENERATE_EMBEDDING are excluded from the generative-inference token quotas because they use different billing structures. Do not combine their token counts with Gemini generation counters when calculating quota headroom.
AI.GENERATE_EMBEDDING can return a statistics JSON value with token_count for text embeddings and token information for some multimodal models. Persist that statistics value with the embedding job, model and input type. Use the applicable embedding pricing documentation and the BigQuery execution cost to build a separate forecast.
Add request-response logging when counters are not enough
Agent Platform request-response logging is optional and must be enabled separately for each foundation model and region. The logs are stored in BigQuery. Google’s documented lookup uses labels.bigquery_job_id_prefix inside the full_request field to find the request and response associated with a job.
Use the correct regional lookup
- For queries in the US, specify
us-central1. - For queries in the EU, specify
europe-west4.
Logging is valuable when you need to explain prompt growth, retries, model changes or row-level failures. It can show the request and response context that a query-level counter cannot. Logging does not replace Cloud Billing reconciliation: it explains activity, while billing remains the authority for posted charges.
Quick Recap
A practical operating workflow
- Inventory functions. Classify each query as generation, embedding or managed/task-specific, and record its model, endpoint and region.
- Measure a sample. Run
AI.COUNT_TOKENSon representative inputs, including long and multimodal cases where applicable. - Execute a controlled job. Capture row count, prompt version, model settings and the BigQuery job ID.
- Record observed counters. From Query results → Job information, save input, output, thought and cache token counts for Gemini generation jobs.
- Record embedding statistics separately. For embedding functions, persist returned
statisticsdata such astoken_countwhen present. - Attribute billing. Filter Vertex AI in Cloud Billing by
bigquery_job_id_prefix, applying the documented length and lowercase rules. - Wait for posting. Recheck after up to 24 hours before closing a reconciliation period.
- Calculate scenarios. Apply current model rates to low, expected and high token volumes, then add BigQuery bytes or slot costs.
- Set guardrails. Use IAM & Admin quota overrides for stricter daily limits and alert before the expected scenario approaches them.
- Investigate variance. Compare model, region, prompt version, row mix, candidates, reasoning settings, cache behavior, retries, failures and BigQuery execution cost.
Checklist for a defensible monthly report
- The report names the function class, model, endpoint and region.
- Input estimates came from representative samples rather than one short test row.
- Completed jobs supplied observed input, output, thought and cache counters where applicable.
- Embedding usage is kept outside the generative token-quota totals.
- The token equation uses the current model’s published rate units and bills only documented categories.
- BigQuery bytes or slot charges are added to endpoint charges.
- Cloud Billing was checked after the possible 24-hour posting delay.
- Low, expected and high scenarios state their row-count and token assumptions.
- Daily quotas are labeled as global, midnight-Pacific guardrails rather than monthly prices.
- Quota failures, retries and partial row failures have an operational response.
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.
