GLM 5.3 Flash
- Input / 1M tokens
- $0.04
- Output / 1M tokens
- $0.08
- Providers
- 37
- Uptime, 24 h
- 100%
API models · Side by side
GLM 5.3 Flash vs GPT-6 Luna
GLM 5.3 Flash and GPT-6 Luna are worth comparing for applications that process many requests. Both offer large context windows, but GLM adds video input and the two have different input-to-output pricing. A short extraction and a long generated report need different budgets.
GPT-6 Luna
- Input / 1M tokens
- $0.10
- Output / 1M tokens
- $0.50
- Providers
- 3
- Uptime, 24 h
- 100%
The short answer
Consider GLM for a workflow that needs video alongside text and images. For text, images or files, compare Luna as well. Use your expected answer length in the calculator: the model with the cheaper prompt is not necessarily cheaper for the whole task.
Estimate your monthly bill
Enter your own usage. We apply each model's current rates, including long-prompt pricing, for a 30-day month.
Try:
GLM 5.3 Flash$3.57per month
GPT-6 Luna$13.50per month
A token is roughly three-quarters of an English word. Cached input, reasoning tokens, tools and media are not included.
Compare the details
Compare context, supported inputs and usage costs. Highlighted rows show a difference.
| Feature | GLM 5.3 Flash | GPT-6 Luna |
|---|---|---|
| Input formats | image, text, video | file, image, text |
| Output formats | text | text |
| Context / position limit | 1,048,576 | 1,050,000 |
| Endpoint output limit (tokens) | 131,072 | 128,000 |
| Developer features | frequency penalty, Reasoning output, logit bias, logprobs, Output limit, min p, parallel tool calls, presence penalty, Reasoning controls, Thinking effort, repetition penalty, Response format, Seed, Stop sequences, Structured output, Temperature, Tool selection, Tool calling, top k, top logprobs, Sampling controls | Reasoning output, Completion limit, Output limit, Reasoning controls, Thinking effort, Response format, Seed, Structured output, Tool selection, Tool calling, Response length |
| Cached input / 1M tokens | $0.03 | $0.01 |
| Cache write / 1M tokens | Unknown | $0.13 |
| USD input / 1M tokens | $0.04 | $0.10 |
| USD output / 1M tokens | $0.08 | $0.50 |
| Available API providers | 37 | 3 |
| Best provider uptime, last 24 h | 100% | 100% |
| Price range across providers (per 1M tokens) | Input $0.040–$0.22, output $0.078–$0.75 across 37 providers | Set by OpenAI |
| Model identifier | z-ai/glm-5.3-flash | openai/gpt-6-luna |
API pricing
Prices are in USD per million tokens. Input is what you send; output is what the model generates. Chat subscriptions and API usage are billed separately.
GPT-6 Luna: long-prompt rates
- Long-prompt threshold: 272,000 prompt tokens. Input: $0.20 / 1M. Output: $0.75 / 1M.
Tools, images and additional reasoning can add charges. The examples below cover input and output tokens only.
Compare local modelsWhat those prices mean in practice
Each example uses the same token budget for both models. These are calculations, not measured workloads. Cache discounts, media, tools and extra reasoning are excluded.
| Workload | GLM 5.3 Flash | GPT-6 Luna |
|---|---|---|
| 1,000 short requestsPer request: 2,000 input + 500 output tokens | $0.12 | $0.45 |
| 100 document summariesPer request: 20,000 input + 1,000 output tokens | $0.09 | $0.25 |
| 10 long-document requestsPer request: 300,000 input + 2,000 output tokens | $0.12 | $0.61 |
Estimate = requests × (input tokens × input price + output tokens × output price) ÷ 1,000,000. Long-prompt rates are applied where the pricing data provides them. Real requests can use different amounts of output.
Beyond the price tag
Start with what arrives in your application
GLM’s endpoint accepts text, images and video; Luna’s accepts text, images and files. That distinction matters for a support tool receiving screen recordings versus one processing uploaded documents. Do not assume every attachment works in both. Establish how your client sends the material and whether it needs preprocessing before committing to either integration.
Extraction and writing have opposite token shapes
An invoice extractor might read thousands of tokens and return a few fields. A writing assistant may receive a short brief and generate pages. The first spends mostly on input; the second spends more on output. Compare those two workloads separately in the calculator. A blended monthly total can hide that one feature is carrying most of the expense.
Luna has a long-prompt billing threshold
Luna’s higher rates start at 272,000 prompt tokens in the pricing used here. GLM has no corresponding long-prompt tier in this comparison’s data. If you send large batches of records together, check whether each request crosses that boundary. Combining records can save repeated instructions, but it can also move a request into a different price tier.
Structured answers need application checks
For a classification or extraction feature, define the fields, allowed values and treatment of missing information. Both endpoints expose structured-response controls, but your application should still validate the returned result. A valid JSON object can contain a wrong invoice total. The useful comparison is how often each model produces a correct, usable record and how much correction that requires.
Keep video costs separate from text examples
The examples on this page use input and output token counts. They do not turn the length of a video into a price. For a recording workflow, establish how frames and any audio are processed by the actual endpoint. A transcription step, frame selection or separate audio model may add spending outside the two text rates shown here.
Common questions
Does Flash mean GLM is faster than Luna?
The name is not a speed measurement. Response time depends on the hosting service, request size and generation settings; this comparison does not claim an independently measured speed winner.
Which should I use for a document workflow?
Check the accepted file format first. Then compare extraction accuracy and the cost of the complete request, including any document parsing your application performs before calling the model.