Qwen3.8 27B API
Updated
Qwen3.8 27B runs on BroadNet Inference as broadnet/qwen3.8-27b through the OpenAI-compatible https://inference.broadnet.ai/v1. It costs $0.175 input, $0.0435 cached input and $2.17 output per 1M tokens, with a 262,144-token context. It needs a qualifying top-up (not the starting credit), and inference content is not retained.
At a glance
- Text
- Vision (image input)
- Tool calling
- Reasoning
- Structured output
- 262,144-token context
- Text and image input
- Tools, structured output and reasoning
- BroadNet test: fluent Arabic writing, Gulf and Egyptian dialects understood, accurate Arabic–English translation
| Model ID | broadnet/qwen3.8-27b |
|---|---|
| Endpoint | /v1/chat/completions |
| Supported inputs | text, image (vision: understanding, not generation) |
| Context | 262,144 tokens, prompt and answer together |
| Input | $0.175 per 1M tokens |
| Cached input | $0.0435 per 1M tokens (reported cache hits only) |
| Output | $2.17 per 1M tokens |
| Pay-as-you-go limits | 30 requests a minute, 4 in flight per key |
| Access | Top-up required: opens after a qualifying top-up; the starting credit does not cover it |
Prices on this page are from BroadNet's price feed snapshot dated , not checked live here. See official pricing for current rates.
official pricingReport a wrong price or number (email to info@broadnet.ai, with this page's link filled in).
What does BroadNet publish about Qwen3.8 27B?
From BroadNet's model catalogue, checked 15 November 2026. Measurements are BroadNet's, on its hardware, unless a public leaderboard is named; we have not reproduced them.
| Serving in BroadNet's lab | GGUF Q4_K_M with a vision projector and MTP drafter on llama.cpp; AWQ INT4 on BroadNet's patched vLLM for AMD ROCm |
|---|---|
| Context on the API | 262,144 tokens (pricing.json) |
BroadNet's measurements
- 55 tokens per second on a single stream with multi-token prediction, and 92% MTP acceptance on ROCm.
How BroadNet chooses and serves models: its GPU hardware and serving approach and the models it evaluated and ruled out.
Published benchmark results
BroadNet did not run these benchmarks; results are as published by the Qwen model card and may use different settings. BroadRouter is BroadNet's site. They refer to the published model Qwen/Qwen3.8-27B; not measured on BroadNet's serving.
| Benchmark | Result | Setting or note | Source |
|---|---|---|---|
| GPQA Diamond | 89.2 | not stated | Qwen model card, checked 2026-11-16 |
| LiveCodeBench v6 | 90.3 | not stated | Qwen model card, checked 2026-11-16 |
| SWE-bench Pro | 61.7 | not stated | Qwen model card, checked 2026-11-16 |
| Terminal Bench 2.1 (Terminus) | 73.0 | not stated | Qwen model card, checked 2026-11-16 |
| IFBench | 79.5 | not stated | Qwen model card, checked 2026-11-16 |
| OSWorld-Verified | 84.3 | not stated | Qwen model card, checked 2026-11-16 |
| OmniDocBench 1.5 | 91.1 | not stated | Qwen model card, checked 2026-11-16 |
Side by side: Qwen3.8 27B vs DeepSeek V4.1 Flash, Qwen3.8 27B vs GLM 5.3 and Qwen3.8 27B vs BroadNet Gemma 4 VL 26B, with prices, context and workload costs. Qwen3.8 27B is also sold by 15 other providers, from $0.039 input and $1.25 output per 1M tokens on their own pricing pages: Qwen3.8 27B API pricing by provider. The same text is a different number of tokens on each model: see how Qwen3.8 27B's tokenizer compares on Arabic, English and code in the tokenizer efficiency dataset.
Call it in Python
Install the OpenAI SDK with pip install openai, export BROADNET_API_KEY, then run:
from openai import OpenAI
import base64, os
client = OpenAI(base_url="https://inference.broadnet.ai/v1", api_key=os.environ["BROADNET_API_KEY"])
image = base64.b64encode(open("invoice.jpg", "rb").read()).decode()
response = client.chat.completions.create(
model="broadnet/qwen3.8-27b",
max_tokens=2000,
messages=[{"role": "user", "content": [
{"type": "text", "text": "What is the total on this invoice?"},
{"type": "image_url", "image_url": {"url": "data:image/jpeg;base64," + image}},
]}],
)
print(response.choices[0].message.content)Or with curl
curl https://inference.broadnet.ai/v1/chat/completions \
-H "Authorization: Bearer $BROADNET_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"broadnet/qwen3.8-27b","messages":[{"role":"user","content":"Reply with one short sentence."}]}'Qwen3.8 27B reasons before it answers. The reasoning counts toward max_tokens and is reported as usage.completion_tokens_details.reasoning_tokens, inside the completion (output) token count, so budget for it as output. Leave 1,000 tokens or more, or set reasoning_effort to low, or the answer can come back empty. Why answers come back empty.
Guides for Qwen3.8 27B
- Use the OpenAI Python SDK with BroadNet Inference
- Use the OpenAI Node.js SDK with BroadNet Inference
- First request to BroadNet Inference with curl
- Stream chat completions and handle dropped connections
- Why a reasoning model returns an empty answer
- Use LangChain with BroadNet Inference
- Use Cline with BroadNet Inference (OpenAI Compatible provider)
- Use Roo Code with BroadNet Inference (OpenAI Compatible)
- Use Continue with BroadNet Inference (config.yaml and apiBase)
- Add BroadNet Inference to opencode as a custom provider
- Connect Open WebUI to BroadNet Inference
- Add BroadNet Inference to LibreChat as a custom endpoint
- Use BroadNet Inference in n8n (OpenAI credential, Base URL)
- Use the Vercel AI SDK with BroadNet Inference
- Route LiteLLM to BroadNet Inference (SDK and proxy)
- Run the OpenAI Agents SDK on BroadNet Inference
- Read Arabic invoices and receipts with a vision model
- Use Zed with BroadNet Inference as an OpenAI-compatible provider
- Use Kilo Code with BroadNet Inference (OpenAI Compatible provider)
- Add BroadNet Inference to goose as a custom provider
- Use Pydantic AI with BroadNet Inference (OpenAICompatibleProvider)
- Use CrewAI with BroadNet Inference as a custom OpenAI endpoint
- Use LlamaIndex with BroadNet Inference through OpenAILike
- Add BroadNet Inference to Dify with the OpenAI-API-compatible plugin
- Use Jan with BroadNet Inference as a custom OpenAI-compatible provider
- Connect SillyTavern to BroadNet Inference (Custom OpenAI-compatible)
- Use Spring AI with BroadNet Inference (OpenAI base-url)
- Call BroadNet Inference from Go with the openai-go SDK
- Build LangGraph agents on BroadNet Inference
- Use Haystack pipelines and agents with BroadNet Inference
- Use Semantic Kernel with BroadNet Inference
- Use Mastra agents with BroadNet Inference
- Get structured output with Instructor on BroadNet Inference
- Use DSPy with BroadNet Inference (set model_type="chat")
- Evaluate BroadNet models with promptfoo
- Run smolagents on BroadNet Inference with OpenAIModel
- Connect Flowise to BroadNet Inference (OpenAI Custom Model)
- Connect Langflow to BroadNet Inference
Use Qwen3.8 27B in the console
The BroadNet console on inference.broadnet.ai manages accounts and keys. It does not receive your choice from this page, so:
- Sign in or create an account.
- Add a qualifying top-up (the starting credit does not cover this model), then create a pay-as-you-go key.
- Send this model ID as
modelto/v1/chat/completions:broadnet/qwen3.8-27b
Open the BroadNet consoleEstimate the costRead the API reference
Compare Qwen3.8 27B
Questions
What is the model ID for Qwen3.8 27B?
Pass broadnet/qwen3.8-27b as model to /v1/chat/completions. The base URL is https://inference.broadnet.ai/v1.
How much does Qwen3.8 27B cost?
$0.175 input, $0.0435 cached input and $2.17 output per 1M tokens, from your account balance.
What is the context window of Qwen3.8 27B?
262,144 tokens per request, prompt and answer together.
Can I use Qwen3.8 27B without a top-up?
No. Qwen3.8 27B opens after a qualifying top-up. The starting credit covers BroadNet Gemma 4 VL 26B, BroadNet Transcribe Arabic, BroadNet Transcribe and Gemma hourly keys.
How does Qwen3.8 27B score on benchmarks?
As published, the Qwen model card reports 89.2 on GPQA Diamond, 90.3 on LiveCodeBench v6 and 61.7 on SWE-bench Pro (checked 2026-11-16). BroadNet did not run these benchmarks, and sources may use different settings, so the figures are not directly comparable. Every published result we found, with its setting and source.
Is my data stored?
Inference content is not retained: prompts, images and answers are processed for the request only. Necessary account, billing and security metadata, such as timings, token counts and cost, is retained as described in BroadNet's privacy statement. Read how.