API Need30% off
Uptime
Total Context
1.0486M
Max Output
128K
Input
$0.98/ 1M
Output
$3.08/ 1M
Cache Read
$0.182/ 1M
Cached Input Storage
Limited-time Free
ZA
Z.AI: GLM 5.2 API
z-ai/glm-5.2
GLM 5.2 API is a text-only model with a 1.05M context and lower official token rates than the premium OpenAI and Anthropic tiers. Its appeal is value-oriented long-context reasoning rather than multimodal input. [1]
The highlighted max run ranks #9 with a 58.27 median score, while the standard run appears at #40. That large mode difference makes reasoning configuration central to any GLM 5.2 API cost and quality comparison. [1][2]
MODALITIES
Price$0.98 / $3.08 / 1MCONTEXT1.0486M
Playground
Test GLM 5.2 API with representative production prompts before using glm-5.2 in a live workflow.
Providers
APINEED lists GLM 5.2 API with text input, text output, and a 1.048576M context window. APINEED routing and prepaid rates are identified separately. [1]
Discount
GLM 5.2 uses a 30% APINEED discount, not the 50% rate shown on several US models. Its official $1.40 input and $4.40 output rates still keep the discounted total competitive. The table compares official GLM 5.2 API pricing with the APINEED prepaid rate.
Official
Provider baselineInput $1.4/ 1MOutput $4.4/ 1M
BaselineAPINEED30% off
Prepaid creditsInput $0.98/ 1MOutput $3.08/ 1M
Availability
APINEED continuously monitors GLM 5.2 API access and keeps requests on healthy capacity.
System statusLast 90 days
GLM 5.2 API Benchmarks
At #9 in max mode, GLM 5.2 places above GPT-5.6 Luna and DeepSeek V4 Pro. The standard #40 row is much lower but faster, illustrating the trade-off between reasoning depth and response time. The complete LLM2014 logic 2026-07 table remains below, with GLM 5.2 API highlighted when a current row exists. [2]
| 1st | GPT-5.5 (xhigh) | 83.80 | 77.46 | 7.57% | 494s | 30,811 | $25.51 | $29.57 |
| 2nd | Kimi-K3 (max) | 82.91 | 74.80 | 9.78% | 1095s | 39,912 | $16.76 | $15.00 |
| 3rd | Claude Opus 4.8 (xhigh) | 82.62 | 66.70 | 19.27% | 791s | 41,833 | $28.86 | $24.64 |
| 4th | GPT-5.6 Sol (xhigh) | 81.94 | 72.99 | 10.92% | 369s | 17,433 | $14.43 | $29.57 |
| 5th | Claude Opus 5 (xhigh) | 78.38 | 71.88 | 8.29% | 426s | 26,392 | $18.21 | $24.64 |
| 6th | Qwen3.7-Max (xhigh) | 74.56 | 66.95 | 10.21% | 374s | 54,207 | $7.81 | $5.14 |
| 7th | GLM-5.2 (max) | 73.68 | 58.27 | 20.91% | 890s | 45,683 | $5.12 | $4.00 |
| 8th | Gemini 3.1 Pro (high) | 73.36 | 58.46 | 20.31% | 214s | 25,905 | $8.58 | $11.83 |
| 9th | GPT-5.6 Luna (xhigh) | 69.49 | 51.97 | 25.21% | 242s | 38,083 | $6.31 | $5.91 |
| 10th | DeepSeek V4 Pro (max) | 68.00 | 49.97 | 26.51% | 1468s | 60,558 | $1.45 | $0.86 |
| 11th | Grok 4.5 (high) | 67.59 | 56.72 | 16.08% | 451s | 43,372 | $7.18 | $5.91 |
| 12th | Muse Spark 1.1 | 66.18 | 57.23 | 13.52% | 542s | 42,416 | $4.98 | $4.19 |
| 13th | Gemini 3.5 Flash (high) | 65.92 | 60.39 | 8.39% | 188s | 36,372 | $9.03 | $8.87 |
| 14th | Doubao-Seed-2.1-pro (high) | 58.69 | 49.35 | 15.91% | 2003s | 85,059 | $10.21 | $4.29 |
| 15th | Qwen3.7-Plus (high) | 58.50 | 44.09 | 24.63% | 1237s | 57,152 | $1.83 | $1.14 |
| 16th | Gemini 3.6 Flash (high) | 56.87 | 41.74 | 26.6% | 206s | 26,329 | $5.45 | $7.39 |
| 17th | Tencent Hy3 (high) | 54.42 | 44.60 | 18.04% | 1035s | 52,037 | $0.83 | $0.57 |
| 18th | Claude Sonnet 5 (xhigh) | 51.37 | 43.14 | 16.02% | 529s | 44,118 | $12.18 | $9.86 |
| 19th | DeepSeek V4 Flash (max) | 50.62 | 36.24 | 28.41% | 611s | 49,593 | $0.40 | $0.29 |
| 20th | MiniMax-M3 | 49.18 | 40.94 | 16.75% | 985s | 56,467 | $1.90 | $1.20 |
| 21st | Claude Opus 5 | 42.08 | 31.73 | 24.6% | 154s | 10,180 | $7.02 | $24.64 |
| 22nd | Claude Opus 4.6 | 36.88 | 25.79 | 30.07% | 65s | 4,387 | $3.03 | $24.64 |
| 23rd | Doubao-Seed-2.0-lite 0428 (high) | 35.32 | 26.77 | 24.21% | 656s | 26,726 | $0.38 | $0.51 |
| 24th | Gemini 3.5 Flash (minimal) | 33.24 | 22.22 | 33.15% | 44s | 6,189 | $1.54 | $8.87 |
| 25th | Ling-3.0-flash | 32.53 | 20.62 | 36.61% | 688s | 86,204 | $0.00 | $0.00 |
| 26th | Gemini 3.5 Flash Lite (high) | 30.23 | 22.82 | 24.51% | 188s | 18,297 | $1.26 | $2.46 |
| 27th | GPT-5.5 Instant | 28.87 | 17.66 | 38.83% | 25s | 1,673 | $1.39 | $29.57 |
| 28th | Gemma 4 31B | 27.91 | 22.04 | 21.03% | 770s | 15,050 | $0.17 | $0.39 |
| 29th | MiMo-V2.5-Pro | 26.91 | 13.00 | 51.69% | 478s | 29,443 | $0.71 | $0.86 |
| 30th | Qwen3.7-Plus | 26.19 | 16.23 | 38.03% | 314s | 8,692 | $0.28 | $1.14 |
| 31st | Step-3.7-Flash | 26.18 | 13.74 | 47.52% | 258s | 44,912 | $1.46 | $1.16 |
| 32nd | Qwen3.5-27B | 24.96 | 17.98 | 27.96% | 451s | 26,391 | $0.51 | $0.69 |
| 33rd | Gemini 3.1 Flash Lite (high) | 23.78 | 14.30 | 39.87% | 82s | 28,351 | $1.17 | $1.48 |
| 34th | Claude Sonnet 5 | 23.64 | 10.86 | 54.06% | 109s | 5,962 | $1.65 | $9.86 |
| 35th | Qwen3.7-Max | 22.45 | 18.53 | 17.46% | 196s | 6,710 | $0.97 | $5.14 |
| 36th | ERNIE 5.1 | 20.49 | 15.49 | 24.4% | 457s | 24,093 | $1.73 | $2.57 |
| 37th | openPangu-2.0-Flash | 19.83 | 10.76 | 45.74% | 571s | 28,199 | $0.18 | $0.23 |
| 38th | LongCat-2.0 | 19.40 | 9.74 | 49.79% | 395s | 14,886 | $0.48 | $1.14 |
| 39th | DeepSeek V4 Flash | 19.39 | 13.37 | 31.05% | 63s | 5,286 | $0.04 | $0.29 |
| 40th | Mistral Medium 3.5 | 18.44 | 13.87 | 24.78% | 194s | 27,854 | $5.77 | $7.39 |
| 41st | Doubao-Seed-2.1-pro | 17.49 | 9.74 | 44.31% | 305s | 6,009 | $0.72 | $4.29 |
| 42nd | GLM-5.2 | 14.80 | 10.43 | 29.53% | 23s | 953 | $0.11 | $4.00 |
| 43rd | Ling-2.6-1T | 13.05 | 8.39 | 35.71% | 229s | 4,770 | $0.31 | $2.29 |
| 44th | iFLYTEK Spark X2 | 12.39 | 6.21 | 49.88% | 458s | 11,285 | $0.09 | $0.29 |
| 45th | Gemini 3.1 Flash Lite | 11.79 | 10.28 | 12.81% | 9s | 773 | $0.03 | $1.48 |
| 46th | Mistral Medium 3.5 | 10.83 | 7.13 | 34.16% | 17s | 2,546 | $0.53 | $7.39 |
| 47th | Ling-2.6-flash | 10.07 | 4.85 | 51.84% | 25s | 4,997 | $0.04 | $0.30 |
Quick Start
Connect GLM 5.2 API without changing the OpenAI-style request shape. Use GLM 5.2 for long text corpora, budget-sensitive reasoning, and asynchronous analysis. Do not send images or files directly because this catalog entry is text input only.
Get your API key
Create an APINEED key for GLM 5.2 API and keep it in an environment variable.
export API_NEED_API_KEY=sk-apineed-v1-...Make your first request
Use glm-5.2 for GLM 5.2 API with the APINEED API. The request shape is compatible with OpenAI chat completions, so most SDKs only need a base URL change.
TypeScript SDKPythoncURLOpenAI SDK
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.API_NEED_API_KEY,
baseURL: "https://apineed.com/v1"
});
const result = await client.chat.completions.create({
model: "z-ai/glm-5.2",
messages: [
{ role: "user", content: "Why is the sky blue?" }
]
});
console.log(result.choices[0].message.content);Enable streaming and fallbacks
Add stream: true when GLM 5.2 API should return server-sent events. APINEED keeps routing, provider health, and fallback handling behind the same endpoint.
curl https://apineed.com/v1/chat/completions \
-H "Authorization: Bearer $API_NEED_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "z-ai/glm-5.2",
"stream": true,
"messages": [{ "role": "user", "content": "Hello" }]
}'GLM 5.2 API Endpoint
GLM 5.2 API accepts chat conversations here with streaming or non-streaming text output; the supported controls are listed on this page.
POST
/v1/chat/completions- Authorization
- Bearer $API_NEED_API_KEY
- Content-Type
- application/json
- HTTP-Referer
- optional - your site URL, for rankings
- X-Title
- optional - your site name, for rankings
- Model
- z-ai/glm-5.2
Parameters
GLM 5.2 API parameters currently listed for z-ai/glm-5.2 by the OpenRouter Models API. [1]
| Name | Type | Status | Description |
|---|---|---|---|
frequency_penalty | number | Supported | Penalizes repeated token frequency. |
include_reasoning | boolean | Supported | Includes reasoning content in the response when available. |
logit_bias | object | Supported | Adjusts the likelihood of selected tokens. |
logprobs | boolean | Supported | Returns token log probabilities. |
max_tokens | integer | Supported | Limits generated output tokens. |
min_p | number | Supported | Applies minimum-probability sampling. |
parallel_tool_calls | boolean | Supported | Allows multiple tool calls in one model turn. |
presence_penalty | number | Supported | Penalizes tokens already present in the output. |
reasoning | object | Supported | Controls reasoning behavior and token allocation. |
reasoning_effort | enum | Supported | Selects the model reasoning effort. |
repetition_penalty | number | Supported | Controls repetition across generated tokens. |
response_format | object | Supported | Requests a specific response format. |
seed | integer | Supported | Requests deterministic sampling when supported by the provider. |
stop | string or array | Supported | Stops generation at the supplied sequence. |
structured_outputs | boolean | Supported | Enables schema-constrained structured output. |
temperature | number | Supported | Controls sampling randomness. |
tool_choice | string or object | Supported | Controls which tool the model may call. |
tools | array | Supported | Defines tools available to the model. |
top_k | integer | Supported | Restricts sampling to the highest-probability tokens. |
top_logprobs | integer | Supported | Sets how many top-token log probabilities are returned. |
top_p | number | Supported | Controls nucleus sampling. |
GLM 5.2 API Q&A
Model-specific answers for teams comparing GLM 5.2 API on capability, benchmark evidence, integration, and cost.
Why does GLM 5.2 API show different benchmark positions?
LLM2014 includes both max and standard runs. The max run ranks #9; the standard run ranks #40, with different token and latency profiles.
What is GLM 5.2 API’s APINEED discount?
APINEED lists a 30% reduction from official rates, producing $0.98 input and $3.08 output per million tokens.
Can GLM 5.2 API process images?
No. This catalog entry accepts text and returns text; image or file inputs should be routed to a multimodal model.
Which workloads fit its 1.05M context for GLM 5.2?
For GLM 5.2 API, large text archives, long reports, code collections, and retrieval bundles can fit, subject to normal token budgeting and output limits.
Which endpoint serves GLM 5.2 API?
Call GLM 5.2 API through https://apineed.com/v1/chat/completions with glm-5.2 as the model value. Existing OpenAI SDK clients usually need only the APINEED base URL and key.
How do I validate GLM 5.2 API before release?
Evaluate GLM 5.2 API on representative prompts, record quality, latency, and token use, then choose reasoning settings and fallbacks from those results.
How to Deploy the GLM 5.2 API on apineed.com
Deploy GLM 5.2 API through APINEED after validating its model-specific trade-offs above. The API key, credit, and endpoint flow stays consistent across the catalog.
Add credits for GLM 5.2 API
Add credits on APINEED before deploying the GLM 5.2 API. Pay-as-you-go billing lets usage start small and scale with production traffic.
Get your API key
Create an APINEED API key for the GLM 5.2 API. The same key can call GLM 5.2 and other AI APIs through apineed.com.
Set the model slug
Use glm-5.2 as the model value when you deploy the GLM 5.2 API. Keep the APINEED base URL at https://apineed.com/v1.
Send a request to GLM 5.2 API
Send chat completions or responses to the GLM 5.2 API from Claude Code, Codex, or any custom agent. The request keeps its OpenAI-compatible shape.