Open in the interactive docs · markdown for agents: https://neuraldeep.tech/docs/chat.md
Text · Chat completion
OpenAI-compatible /v1/chat/completions. Supports streaming, tools, reasoning (gpt-oss passes reasoning tags through inline). Turns of one conversation stay on one worker automatically.
curl https://api.neuraldeep.tech/v1/chat/completions \
-H "Authorization: Bearer $YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-oss-120b",
"messages": [
{"role":"system","content":"You are helpful."},
{"role":"user","content":"Hello"}
],
"max_tokens": 500,
"temperature": 0.2
}'
from openai import OpenAI
client = OpenAI(api_key="$YOUR_KEY", base_url="https://api.neuraldeep.tech/v1")
r = client.chat.completions.create(
model="gpt-oss-120b",
messages=[{"role":"user","content":"Hello"}],
max_tokens=500,
)
print(r.choices[0].message.content)
On our qwen lanes the turns of one conversation (same key, same opening of the chat) land on the same worker on their own, so the prefix comes from the cache and multi-turn is faster. The
userfield does not affect routing. If you run several parallel copies of the same conversation and want them spread across workers, pass your ownprompt_cache_key.
Reasoning per model. Reasoning is on by default; reasoning_content arrives as a separate field.
gpt-oss-120b: depth is set byreasoning_effort—low/medium/high.qwen3.8-27b: depth is set byreasoning_effort—low/medium/xhigh(defaultlow). The model does not knowhighorminimal; such a value is dropped andlowstays.qwen3.6-35b-a3b: no effort levels,reasoning_efforthas no effect.- To turn thinking off on qwen, use the
-noreasonmodel (qwen3.8-27b-noreason,qwen3.6-35b-a3b-noreason).enable_thinking: falseon the base name has no effect: the gateway drops it and the model still thinks. thinking_token_budgetdoes not work on qwen, neither insidechat_template_kwargsnor at the top level.
Reference sections (markdown for agents)
- Start · key, errors, quotas
- Text · chat
- Vectorization
- Reranking
- Transcription
- Image recognition
- Structured output
- Clients · agents
- Remaining limits · Limits API
- Search · Search API
- OCR · documents
- Images · Image API
- Music · Music API
- PII anonymization · PII Guard
- SpeechCore · STT
- Speech synthesis · TTS API
- Who spoke when · diarization
- Text decisions · FRIDA-Decisions, clef-flash
- Drift · overview
- Files + sandbox preview
- Your own tools, skills, MCP
- Tasks · proactive
- Memory · memory
- Isolation and security
- Agent settings
- Limits · errors
- Book · search and MCP
- Streaming
- SDK · Python / JS
- Privacy