Qwen3.6-27B-FP8

Overview

Qwen3.6-27B-FP8 is a 27-billion-parameter causal language model with a vision encoder, optimized for agentic coding and vision-language tasks. It is hosted on gigarouter as a managed, OpenAI-compatible API, eliminating the need for local infrastructure.

Key Strengths

• Agentic Coding: Enhanced fluency and precision in frontend workflows and repository-level reasoning.
• Thinking Preservation: Retains reasoning context from historical messages to streamline iterative development.
• Extended Context: Native context length of 262,144 tokens, extensible up to 1,010,000 tokens.
• FP8 Quantization: Fine-grained FP8 quantization (block size 128) delivers performance nearly identical to the original model.

Benchmark Results

Qwen3.6-27B achieves competitive scores on coding agent and knowledge benchmarks:

Model Architecture

• Parameters: 27B
• Hidden Dimension: 5120
• Layers: 64
• Hidden Layout: 16 × (3 × (Gated DeltaNet → FFN) → 1 × (Gated Attention → FFN))
• Context Length: 262,144 native, up to 1,010,000 tokens
• Training Stage: Pre-training & Post-training

Best For

Developers building agentic coding applications, particularly those requiring repository-level reasoning, frontend workflow automation, and long-context understanding. The model is also suited for vision-language tasks via its integrated vision encoder.

For further details, see the Qwen3.6-27B blog post.