Which Xploded model to use
Five uncensored models that all call tools. Which one to point your agent at, which one to use for hard problems, for images and for high volume.
Start with GLM 5.3 Flash Xploded
For a coding agent, GLM 5.3 Flash Xploded is the default, and it is what the chat opens on. It has a million tokens of context, calls tools, reads images, and costs a fraction of the full GLM 5.3. Most people never need to change it.
Hard problems: GLM 5.3 Xploded
GLM 5.3 Xploded is the strongest of the five. Artificial Analysis scored it 45 on its intelligence index when we checked on 2 October 2026, against 42 for the Flash model. Reach for it on long reasoning, large refactors and anything the Flash model gets wrong twice. It costs more per token, so it suits the hard steps more than every step.
Images: Flash or MiMo
GLM 5.3 Flash Xploded and MiMo V2.6 Flash Xploded both take images. MiMo is the abliterated one of the line, with the refusal direction removed from its weights rather than retrained away, and it also has a million tokens of context. If one of them balks at a task, try the other; the two were built differently and fail in different places.
Long text-only work: DeepSeek V4.1 Flash Xploded
DeepSeek V4.1 Flash Xploded writes up to 393,216 tokens in one reply, the longest on the line, and scored 39 on the same index. It is text only. Reach for it for long reports, large generated files and long agent sessions that re-read their history; its reasoning effort can be set per request, or turned off for bulk work. It is compared head to head with GLM 5.3 Flash and MiMo V2.6 Flash.
Simple, repetitive steps: Qwen3.8 27B Xploded
Qwen3.8 27B Xploded is the line's 27B model, with working tool calls. It scored 34 on the same index, so give it the simple, repetitive agent steps and keep the hard thinking for GLM.
Switching costs nothing
All five sit behind the same endpoint and the same key, so changing model means changing one id. In Claude Code that is ANTHROPIC_MODEL; remember to set CLAUDE_CODE_MAX_CONTEXT_TOKENS to the new model's context. Current prices for each are on pricing, read from the live catalogue.