Show off your agentic readiness — the badge auto-updates when your score changes.
llms.txt Content
# Deepent Research
> Deepent Research is a refusal-free ("uncensored") LLM chat service. Standard AI assistants stop and refuse questions; Deepent answers them to the end. Powered by abliterated open-weight models (Qwen3.5 / Qwen3.8-27B-abliterated family).
## Agent API (pay-per-request, x402)
Agents can call our inference API without any signup. Payment is handled with the
x402 (HTTP 402) standard in USDC on Base. Price is per request; test the negotiation
with a plain POST.
- Endpoint: POST https://api.deepentresearch.com/v1/agent/chat
- Payment: x402 "exact" scheme, USDC on Base (eip155:8453)
- Price: standard $0.05 / performance $0.15 per request
- Machine-readable manifest: https://deepentresearch.com/.well-known/agent-api.json
- Behavior: returns 402 with a PAYMENT-REQUIRED header describing the payment. Retry with a signed payment header to receive a JSON completion.
Request body: { "messages": [{ "role": "user", "content": "..." }], "model": "standard" }
## MCP server
Agents can also call Deepent as an MCP tool (Streamable HTTP, JSON-RPC 2.0):
- Endpoint: POST https://deepentresearch.com/mcp
- Tools: `deepent_chat` (requires a subscription API key `dp_...` — pass it as the `api_key` argument or an `Authorization: Bearer` header) and `deepent_service_info` (free)
## Tools
- [GGUF VRAM calculator](https://deepentresearch.com/journal/tools/gguf-vram) (ja) / https://deepentresearch.com/en/journal/tools/gguf-vram (en): which GGUF quantization fits a given model size, context length and memory budget. The formula (weights + KV cache + runtime reserve) and the bits-per-weight table are published on the page, so the numbers can be checked.
## Articles (journal)
- [検閲なし AI で文章を書くには?小説・レポートを止められずに書く手順](https://deepentresearch.com/journal/writing-with-uncensored-ai): 場面が激しい小説や、専門的なレポートの下書きで「お答えできません」と止まることがありま
OpenAPI Spec (preview)
openapi: 3.1.0
info:
title: Deepent Research Agent API
version: "1.0"
description: >-
Refusal-free uncensored LLM chat for agents. Pay-per-request with x402
(USDC on Base). No signup or API key: an unpaid request returns HTTP 402
with a PAYMENT-REQUIRED header describing the exact price; retry with a
signed payment header to receive the completion. Standard $0.05 /
performance $0.15 per request.
servers:
- url: https://api.deepentresearch.com
paths:
/v1