Do these 3 things before closing this tab:
1Fix the driver behind crashes, sound loss and screen glitches2Clear out junk files and repair common Windows errors3Scan for outdated or missing drivers - takes under a minuteTo diagnose a token spike, compare actual usage records for the same time period and scope, then trace the increase to individual model requests and agent steps. Answer length alone is not a reliable measure: one task may trigger several model calls, and usage can include input context, generated output, reasoning, tool-call turns, retries, or delegated work. The field names and tracing features below are specific to the documented OpenAI APIs and Agents SDK; other providers may report usage differently.
1. Confirm the spike in usage records
First make sure you are comparing the same accounting interval, project or organization, model, and task scope as the reported cost or quota change. Use the API response fields rather than estimating from the length of the assistant’s visible answer. OpenAI documents these usage fields:
As an Amazon Associate I earn from qualifying purchases.
| API | Input usage field | Output usage field | Total |
|---|---|---|---|
| Chat Completions | usage.prompt_tokens |
usage.completion_tokens |
usage.total_tokens |
| Responses | usage.input_tokens |
usage.output_tokens |
usage.total_tokens |
Depending on the endpoint and model, usage details can also include cached input and reasoning tokens. Those categories help explain where tokens went, but the total is the starting point for verifying whether usage actually rose. See OpenAI’s API usage documentation.
2. Capture usage for every request, not just every task
A single user-facing task can require multiple model requests. A run total is useful for measuring overall impact, but per-request records reveal whether the increase came from more calls, larger calls, or both. Record at least:
#1 Best Overall
- EVOLUTION AMD RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
- A run or task identifier and request identifier
- Model and timestamp
- Request count and input, output, and total tokens
- Cached-input, cache-write, and reasoning details when the API exposes them
The OpenAI Agents SDK documents run-level input, output, and total usage, request counts, and per-request usage entries. Its run totals aggregate model calls made during the run, including calls that produce tool calls or handoffs. See Agents SDK usage.
3. Trace the run step by step
Once you have identified the runs with unusually high totals, inspect their traces and locate the model requests responsible. Correlate each request with the surrounding tool calls, handoffs, retries, and subagent activity. Do not assume a tool itself consumed the tokens: inspect the model turns before and after it, as well as any tool output sent back into the conversation.
OpenAI’s Agents API guidance identifies several possible contributors to usage: instructions, tool definitions, conversation history, user input, files or images, tool results, generated output, reasoning, subagents, retries, and possible cache-write charges. Treat these as places to investigate, not as proof of a cause; confirm each explanation in the request payload and trace. See the Agents API guide and tracing documentation.
Quick wins for a faster PC:
Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Rank #2
- Built for Local AI Development: AMD Ryzen AI Halo is designed for local AI development and inference, featuring 128GB unified memory and support for up to 200B parameter models to build and run intensive AI workloads locally.
- 128GB Unified Memory: Features 128GB LPDDR5x unified memory at 8000 MT/s with 256 GB/s memory bandwidth, providing a shared memory pool across the CPU, GPU, and NPU to support larger AI models.
- AMD Ryzen AI Max+ 395 Processor: Features 16 cores, 32 threads, and Zen 5 architecture, paired with AMD Radeon 8060S integrated graphics featuring 40 RDNA 3.5 compute units and an AMD XDNA 2 NPU with up to 50 TOPS.
- Linux AI Developer Platform: Purpose-built for Linux-based AI development with full AMD ROCm software support and preloaded tools, models, and workflows optimized for local AI development.
- Compact, Connected Design: Includes a 2TB M.2 SSD, 10GbE LAN, Wi-Fi 7, Bluetooth 5.4, USB-C connectivity, and HDMI 2.1b.
4. Separate input growth from output growth
If input tokens increased
Compare the context sent with each request. Look for repeated instructions or conversation history, expanded tool definitions, large file or image inputs, and tool results that are bigger than before. A common diagnostic question is whether the application resends context on each turn or includes an unnecessarily large tool payload. Check the actual request and trace before attributing the spike to any one item.
If output or reasoning tokens increased
Inspect generated text and tool-call arguments, then check reasoning-token details where separately exposed. A short visible response does not necessarily mean the model produced few total output tokens if the endpoint reports reasoning separately. Compare the request-level breakdown with a similar baseline rather than inferring token counts from what the user sees.
5. Interpret cache usage correctly
Where supported, compare cached input, uncached input, and cache-write counts. Cached input is still billed, according to OpenAI’s Agents API guidance, so a high cached-input share alone does not establish that the task is inexpensive. If reuse appears lower than expected, compare the request with a recent baseline and use cache diagnostics where available.
Rank #3
- EVOLUTION RYZEN AI MAX+ 395 MINI PC - GMKtec EVO-X2 is the next evolution in AI mini PC Ryzen Strix Halo series. Thanks to AMD Simultaneous Multithreading (SMT) the core-count is effectively doubled, to 32 threads. Ryzen AI Max+ 395 has 64 MB of L3 cache and can boost up to 5.1 GHz, depending on the workload. The Ryzen AI Max+ 395 is currently rated as the "most powerful x86 APU" on the market for AI computing.
- AI NPU with XDNA 2 ARCHITECTURE - Powered by 16 “Zen 5” CPU cores, 50+ peak AI TOPS XDNA 2 NPU and a truly massive integrated GPU driven by 40 AMD RDNA 3.5 CUs, the Ryzen AI MAX+ 395 is a transformative upgrade and delivers a significant performance boost over the competition. The Ryzen AI Max+ 395 excels in consumer AI workloads like the llama.cpp-powered application: LM Studio. Shaping up to be the must-have app for client LLM workloads, LM Studio allows users to locally run the latest language model without any technical knowledge required and unleash their creativity and productivity.
- AMD RADEON 8090S iGPU GAMING PC - The AMD Radeon RX 8060S offers all 40 CUs with up to 2.9 GHz graphics clock and uses the new RDNA 3.5 architecture. The powerful iGPU is positioned between an RTX 4060 and 4070 laptop GPU and therefore enables gaming in FHD at maximum details in most demanding games. The 8060S can also utilize the full 128GB pool, which is perfect for running LLMs such as Deepseek 70B Q8, which runs comfortably on this machine.
- EIGHT CHANNEL LPDDR5X - LPDDR5X is a new ground breaking memory small form factor installed on-board. With blazing speeds up to to 8000MT/s, it runs 1.5x faster than the DDR5 SODIMMs; 90% better performance over DDR5 SODIMMs in video conferencing and photo editing; 30% better performance in productivity apps; 12% better performance in digital content workloads.
- QUAD SCREEN 8K DISPLAY SUPPORT - EVO-X2 AI Mini PC support 4-screen 4K/8K output via HDMI 2.1 (8K@60Hz), DisplayPort 1.4 (4K@60Hz), and dual USB 4 40Gbps Transfer speed (supporting PD3.0/DP1.4/DATA). Ideal for gaming, video editing, and multitasking, it provides expansive and crisp multi-display support.
OpenAI’s documented prompt-cache diagnostics require an exact prompt prefix and compatible settings, including model, service tier, and tools. The documentation states availability for the Responses API on GPT-5.6 and later supported models; check the current documentation for the model and endpoint you use. See prompt caching and prompt-cache diagnostics.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →6. Treat missing usage as unknown, not zero
Check whether collection is complete before concluding that a run used fewer tokens than expected. With streamed Chat Completions, request usage by setting stream_options: {"include_usage": true}. OpenAI says the usage chunk arrives before [DONE]; if the stream is interrupted, that chunk may be missing. Its absence is not evidence of zero usage.
In Agents API reporting, usage may be null or may change as accounting arrives. A blank or null trace value means the count is unknown at that point, not zero, and recorded counts are not necessarily the final bill. Resolve incomplete records before comparing totals. See streaming usage guidance and the Agents API tracing documentation.
7. Compare equivalent runs, then change one thing
Use similar tasks, the same model and configuration, and the same accounting interval. Separate request-count growth from tokens-per-request growth, then compare the categories that explain each difference:
- Input versus output and reasoning tokens
- Total model requests and tokens per request
- Usage associated with model turns, tool loops, retries, handoffs, and subagents
- Cached versus uncached input and cache writes, when available
- Whether the usage records are complete and settled
After identifying the step or category that grew, change one relevant factor—such as supplied context, tool-result size, loop limits, delegated work, or a cache-sensitive request setting—and compare the resulting traces and usage. The documented measurement surfaces do not establish a universal threshold for what counts as “high”; your useful baseline is a comparable run with the same task and configuration.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallCrashes, 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 minuteQuick Recap
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.
