Inception: Mercury 3 vs Z.ai: GLM 4.7 Flash
A side-by-side comparison of Inception: Mercury 3 ($0.35 / 1M tokens) and Z.ai: GLM 4.7 Flash ($0.21 / 1M tokens) using current USD prices and capabilities from the api14 catalog. Both run through the same OpenAI-compatible API.
Specifications
| Inception: Mercury 3 | Z.ai: GLM 4.7 Flash | |
|---|---|---|
| Provider | Inception | Z.ai |
| Input | $0.35 / 1M tokens | $0.21 / 1M tokens |
| Output | $1.22 / 1M tokens | $1.40 / 1M tokens |
| Cache read | $0.04 / 1M tokens | - |
| Context window | 1,000,000 | 200,000 |
| Max output | 131,072 | 117,964 |
| Input | text | text |
| Output | text | text |
| Tool calling | ||
| Structured output | ||
| Reasoning control | ||
| Added | Oct 16, 2026 | Jan 19, 2026 |
Key differences
- Z.ai: GLM 4.7 Flash is 40% cheaper on input tokens.
- Inception: Mercury 3 is 13% cheaper on output tokens.
- Inception: Mercury 3 reads 5.0× more context (1,000,000 tokens).
- A request with 1,000 input and 500 output tokens costs $0.000963 with Inception: Mercury 3 and $0.000912 with Z.ai: GLM 4.7 Flash.
In depth
Inception: Mercury 3 is a text model from Inception, and Z.ai: GLM 4.7 Flash is a text model from Z.ai. Both are served through api14 as chat models in the text category, meaning they take text in and return text out. Neither one accepts images, audio or other non-text inputs, so if your workload involves anything beyond text you will need a different model on either side. The two differ mainly in scale and in how much control they give you over generation.
The clearest split is context. Mercury 3 carries a context window of 1,000,000 tokens, while GLM 4.7 Flash carries 200,000. That is a fivefold difference, and it matters if you are feeding in long documents, large codebases or extended conversation histories in a single call. On price, the two are not aligned: GLM 4.7 Flash is the cheaper option for input tokens, while Mercury 3 is the cheaper option for output tokens. So the cost question depends on your ratio of prompt tokens to completion tokens. A workload that reads a lot and writes a little leans toward GLM 4.7 Flash, while one that reads a little and writes a lot leans toward Mercury 3. Both expose tools, structured outputs and reasoning controls, so neither has an exclusive feature there.
For practical selection, start with the shape of the request. If you need to pass very large inputs, such as a full repository, a long transcript or a stack of reference documents, Mercury 3 is the natural fit because of its larger context window. If your prompts are moderate and you generate long outputs, such as drafting, summarisation at length or code generation with substantial completions, Mercury 3's cheaper output pricing works in its favour. If your prompts are large but your answers are short, such as classification, extraction or routing over long inputs, GLM 4.7 Flash is the better economic choice. GLM 4.7 Flash also exposes a wider set of sampling parameters, including frequency and presence penalties, logit bias, logprobs, min_p, repetition penalty, seed, top_k, top_p and top_logprobs. If you rely on seeded reproducibility, token-level log probabilities or fine-grained sampling control, that parameter surface is a real reason to pick it. Mercury 3's parameter list is narrower and centred on reasoning effort, response format, stop sequences, temperature and tool choice.
Calling either model through api14 works the same way, because the API is OpenAI-compatible. You point your client at the api14 base URL, send a standard chat completions request, and set the model field to the id you want. To use Mercury 3, set the model to "inception/mercury-3". To use GLM 4.7 Flash, set the model to "z-ai/glm-4.7-flash". Everything else in the request body, the messages array, the tools, the response format, stays the same. Switching between the two is a one-line change, which makes it cheap to benchmark both against your own traffic rather than guessing from specifications.
There are still cases where GLM 4.7 Flash is the better pick even when context is not the constraint. If your task depends on logprobs or top_logprobs for confidence estimation, or on a fixed seed for reproducible sampling, Mercury 3 does not list those parameters, so GLM 4.7 Flash is the one to call. The same applies if you need penalties or top_k and top_p tuning as part of your prompt strategy. Conversely, if you need a very large context and cheap output, Mercury 3 is the one to reach for.
Switch with one string
Both models use the same endpoint and request shape. Change the model value from "inception/mercury-3" to "z-ai/glm-4.7-flash" and prompts, streaming and tools keep working.
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api14.com/api/v1",
apiKey: process.env.API14_API_KEY,
});
const result = await client.chat.completions.create({
"model": "z-ai/glm-4.7-flash",
"messages": [
{
"role": "user",
"content": "Explain vector databases in two sentences."
}
]
});
console.log(result.choices[0].message.content);
More comparisons
- LiquidAI: LFM3-1.7B vs Ricursive: Ripple 65B Preview
- LiquidAI: LFM3-1.7B vs Crouton Alpha
- LiquidAI: LFM3-1.7B vs Tencent: Hy4
- LiquidAI: LFM3-1.7B vs Z.ai: GLM 5.3 FlashX Legal
- Ricursive: Ripple 65B Preview vs Crouton Alpha
- Ricursive: Ripple 65B Preview vs Tencent: Hy4
- Ricursive: Ripple 65B Preview vs Z.ai: GLM 5.3 FlashX Legal
- Crouton Alpha vs Tencent: Hy4