Unquant

Opening your workspace…

Unquant

Qwen 3.8 Flash Next

Our 125B sparse MoE with just 6B active, uncensored at the tensor level: a fast, long-context all-rounder for coding, agents, research and creative work, with vision, tool calling, thinking and up to 262K context.

A fast all-rounder for coding, agents, research and long-context work, and a natural fit for creative writing and roleplay. Works with any client that accepts a custom OpenAI-compatible base URL and model ID.

At a glance

  • Model family: Qwen 3.8 Flash Next
  • Model ID: qwen-3.8-flash-next
  • Plans from $49 / month
  • Usage: unlimited tokens & messages
  • Context: 65,536 tokens on Roleplay, 131,072 on Plus and 262,144 on Standard, Pro, Ultra and Dedicated
  • Uncensored at the tensor level, with vision and tool calling
  • Access: private beta · by invitation

Qwen 3.8 Flash Next subscriptions

  • Qwen Roleplay: $49/month · unlimited tokens & messages · 64K context · 1 concurrent stream
  • Qwen Plus: $69/month · unlimited tokens & messages · 128K context · 1 concurrent stream
  • Qwen Standard: $89/month · unlimited tokens & messages · 262K context · 1 concurrent stream
  • Qwen Pro: $169/month · unlimited tokens & messages · 262K context · 2 concurrent streams
  • Qwen Ultra: $319/month · unlimited tokens & messages · 262K context · 4 concurrent streams
  • Qwen Dedicated: $1,999/month · unlimited tokens & messages · 262K context · 8 concurrent streams

Compare plans · Request beta access · Read the docs

What Qwen 3.8 Flash Next is built for

  • Roleplay and fiction: Long sessions in SillyTavern or Janitor AI.
  • Everyday coding: Quick tool-calling turns for agents and scripts, at low cost.
  • Long-context work: Research, summaries and chat over large documents.
  • High volume: Unlimited messages from the lowest tier up.

Specifications

  • Parameters: 125B (6B active per token)
  • Context: 64K–262K (Tokens, by plan)
  • Input: Text + images (Vision built in)
  • Reasoning: 3 levels (none · low · high)
  • Model ID: qwen-3.8-flash-next
  • Base model: Qwen 3.8 Flash Next by the Qwen team, open weights
  • Attention: Hybrid sparse attention, built for long context
  • API: OpenAI and Anthropic compatible
  • Tool calling, Vision, Streaming, JSON output, OpenAI & Anthropic APIs, Uncensored in the weights, Unlimited tokens

Connect in a minute

Any OpenAI-compatible client works. In Python:

from openai import OpenAI

client = OpenAI(base_url="https://api.unquant.io/v1", api_key="YOUR_UNQUANT_KEY")
reply = client.chat.completions.create(
    model="qwen-3.8-flash-next",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(reply.choices[0].message.content)

Setup guides: SillyTavern · Janitor AI · Open WebUI · Claude Code · API docs

Qwen 3.8 Flash Next FAQ

Is Qwen 3.8 Flash Next uncensored?

Yes. It is uncensored at the tensor level: refusals are removed in the weights rather than bypassed with a prompt. Vision, tool calling and thinking stay intact.

Is it really unlimited?

Yes. Every plan includes unlimited tokens and messages with no caps or meters, from the Roleplay tier up.

How much context do I get?

65,536 tokens on Roleplay, 131,072 on Plus and the full 262,144 on Standard, Pro, Ultra and Dedicated.

Can I control the reasoning?

Yes. Send reasoning_effort "none" for direct answers, or "low" or "high" for more thinking.

Is it good for roleplay?

Yes. It is quick at long context and inexpensive, which suits long-running roleplay in SillyTavern or Janitor AI.