Abliterated is Abliteration's hosted version of Z.ai's open-weight GLM-5.3 with its refusal behavior removed through abliteration, which finds the activation directions that produce refusals and removes them from the weights. It keeps GLM-5.3's coding and agentic strengths, is text-only, and has a 1M-token context window. Reasoning always runs and cannot be turned off (effort: low, high, or max). It supports tool calling, JSON mode and JSON Schema output, streaming, and web search, through OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages request shapes. It is US-hosted in FP8 with zero prompt retention. Image and video input are rejected.
FastRouter routes your requests to this provider. You can also bring your own key.
abliterationRequest parameters you can send with this model.
Median throughput and time to first token per provider over the last week.
OpenAI-compatible. Point your SDK at FastRouter, or call the API directly.
<FASTROUTER_API_KEY> with your key · Get your key →Installz-ai/glm-5.3-flashGLM-5.3-Flash is Z.ai’s flagship natively multimodal GLM-5-series model with a 1M-token context window, accepting text plus images and video and producing text outputs. It targets fast, high-quality general and coding assistance with strong reasoning at “flash” latency and is released as open weights under the MIT license.
z-ai/glm-5.3-flash-abliteratedGLM-5.3 Flash Abliterated is Refuseless's hosted version of Z.ai's open-weight GLM-5.3-Flash with its refusal behavior removed through abliteration. It accepts text and images and returns text, with a 1M-token context window and up to 131,072 output tokens. Reasoning always runs and cannot be turned off (effort: low, high, or max; default low). It supports tool calling and structured output through an OpenAI-compatible chat completions API. Reasoning text is returned in reasoning_content.
z-ai/glm-5.3-flashxGLM-5.3-FlashX is Z.ai's high-speed serving variant of GLM-5.3-Flash, delivering inference speeds of up to 200 tokens/s for faster responses. It inherits Flash's 320B-total/18B-active MoE design, native multimodality, hybrid sparse and linear attention architecture, and 1M-token context window. FlashX runs on a purpose-built serving stack (SGLang-based inference engine, W8A8 quantization, mixed INT8/FP8/BF16 cache quantization, and encode-prefill-decode disaggregation), reporting roughly 3x end-to-end throughput over Z.ai's initial baseline on the same hardware. It is suited for interactive coding agents, browser/computer-use loops, visual coding, and other multi-step workflows where latency compounds.
Model scores sourced from ArtificialAnalysis