xingchen-agi · Open weights
Xing 4.0 29B A4B
China Telecom’s Xing 4.0 is an Apache 2.0 open-weight mixture-of-experts model with 29B total and 4B active parameters, designed for coding, reasoning and tool use. Supports text input, structured output and a shared 262,144-token input/output context. This route does not use a TEE.
- Reasoning
- Tools
- Structured output
- Parallel tools
- Input
- Text
- Output
- Text
- Reasoning effort
- none · minimal · low · medium · high · xhigh · max
- Discovery tier
- Open weights
Pricing
USD per 1M tokens- Input
- $0.22
- Output
- $0.90
- Cache read
- $0.011
Reference prices as published by the model provider. API example
Specifications
- Context window
- 262.1K
- Max output
- 262.1K
- Average output (7 days)
- 664 tokens
- Longer than 59% of 473 models
Side by side
Xing 4.0 29B A4B compared
Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.
| Xing 4.0 29B A4Bxingchen-agiThis model | MiniMax M3MiniMaxRemove | |
|---|---|---|
| Provider | xingchen-agi | MiniMax |
| Discovery tier | Open weights | Open weights |
| Added | Oct 24, 2026 | Jun 1, 2026 |
| Limits | ||
| Context window | 262.14K | 512KBest |
| Max output | 262.14KBest | 80K |
| Average output | 664 tokens | 491 tokens |
| Pricing, USD per 1M tokens | ||
| Input | $0.22Best | $0.23 |
| Output | $0.90Best | $0.96 |
| Cache read | $0.011Best | $0.05 |
| Performance | ||
| Throughput | — | 114.0 tok/s |
| Time to first token | — | 3.2 s |
| Request success (24h) | 40.6% | 99.9%Best |
| Benchmark indexes | ||
| Intelligence Index | — | 29.2 |
| Coding Index | — | 58.6 |
| Capabilities | ||
| Reasoning | Yes | No |
| Tools | Yes | Yes |
| Vision | No | Yes |
| Structured output | Yes | Yes |
| Parallel tools | Yes | No |
| Reasoning effort | none, minimal, low, medium, high, xhigh, max | — |
Public evaluations
Benchmarks
Independent evaluation results, with how each score ranks among the models in the Noumen catalog.
- Xing 4.0 29B A4B
- MiniMax M3
Intelligence Index
- Xing 4.0 29B A4B—
- MiniMax M329.2
Indexes
Coding Index
Composite of the coding evaluations
- Xing 4.0 29B A4B: —
- MiniMax M3: 58.6
Agentic work
AutomationBench-AA
Workflow automation, penalized for guardrail violations
- Xing 4.0 29B A4B: —
- MiniMax M3: 21.3%
AutomationBench-AA tasks completed
Workflows finished end to end without a guardrail violation
- Xing 4.0 29B A4B: —
- MiniMax M3: 4.4%
Harvey LAB-AAv1.1
Legal tasks passed in full, with no material hallucination
- Xing 4.0 29B A4B: —
- MiniMax M3: 0.0%
AA-Briefcasev1.1
Agentic knowledge work, rated head to head
- Xing 4.0 29B A4B: —
- MiniMax M3: 1,089 Elo
GDPval-AAv2.1
Economically valuable occupational tasks, rated head to head
- Xing 4.0 29B A4B: —
- MiniMax M3: 1,243 Elo
Document reasoning
GDP.pdf
Professional reasoning over PDF documents, all-pass rate
- Xing 4.0 29B A4B: —
- MiniMax M3: 9.8%
AA-LCRv1.1
Reasoning across long documents
- Xing 4.0 29B A4B: —
- MiniMax M3: 83.0%
MLCR-AA
Long-context reasoning over medical records
- Xing 4.0 29B A4B: —
- MiniMax M3: 4.4%
Reasoning and knowledge
Humanity’s Last Exam
Expert-level questions across academic fields
- Xing 4.0 29B A4B: —
- MiniMax M3: 39.0%
GPQA Diamond
Graduate-level science questions
- Xing 4.0 29B A4B: —
- MiniMax M3: 92.9%
IFBench
Following precise output instructions
- Xing 4.0 29B A4B: —
- MiniMax M3: 82.9%
Coding
Terminal-Benchv4.0
Practical software engineering in a terminal
- Xing 4.0 29B A4B: —
- MiniMax M3: 2.0%
SciCode
Scientific computing in Python
- Xing 4.0 29B A4B: —
- MiniMax M3: 47.1%
Legacy evaluations
Terminal-Bench Hard
Earlier terminal task suite
- Xing 4.0 29B A4B: —
- MiniMax M3: 42.4%
τ²-Bench Telecom
Tool use in a customer-support simulation
- Xing 4.0 29B A4B: —
- MiniMax M3: 88.9%
AA-LCR (unversioned)
Long-context reasoning, earlier grading
- Xing 4.0 29B A4B: —
- MiniMax M3: 83.0%
Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis
- Xing 4.0 29B A4B
- MiniMax M3
| Leaderboard | Xing 4.0 29B A4B | MiniMax M3 |
|---|---|---|
| GPQA Diamond | Not in top 20 | #8 · 90.9% |
Ranks among Noumen catalog models on Epoch AI leaderboards. Only top-20 placements are listed.Epoch AI
- Xing 4.0 29B A4B
- MiniMax M3
| Leaderboard | Xing 4.0 29B A4B | MiniMax M3 |
|---|---|---|
| Agentic Game Dev | Not in top 20 | #16 · 1124 |
| Full Stack | Not in top 20 | #18 · 1197 |
| Native Apps | Not in top 20 | #17 · 1162 |
Ranks among Noumen catalog models on Design Arena leaderboards. Only top-20 placements are listed.Design Arena
Live routing data
Performance and reliability
Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.
- Prompt cache hits
- 43.0%
- About $0.13 per 1M input tokens with caching
- Request success (24h)
- 40.9%
- Failing
Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.
- Healthy (98% or more)
- Degraded (90–98%)
- Down (under 90%)
- No requests
Developers
Use it through the API
Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.
curl https://www.noumen.xyz/api/v1/chat/completions \
-H "Authorization: Bearer $NOUMEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "xingchen-agi/xing4.0-29b-a4b",
"reasoning_effort": "medium",
"messages": [{ "role": "user", "content": "Explain this stack trace." }]
}'import os
from openai import OpenAI
client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
model="xingchen-agi/xing4.0-29b-a4b",
messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)