stealthmodels.

MODEL BENCHMARKS

GLM-5.3-Flash: Raccoon

Formerly Ox Alpha, this MIT-licensed multimodal MoE combines a million-token context with low hosted token rates. Read the Ox Alpha reveal and model guide →

Explore the results Specs and pricing
1 model1 resultsHow StealthMark works
StealthMark48.9/100Overall benchmark scoreGeneral Intelligence42.3/100Spatial & physical reasoningVisual Intelligence59.7/100Visual construction & detail

DATASHEET

GLM-5.3-Flash specs and pricing

A good fit forCost-conscious coding and multimodal agents through an API, or self-hosting where several hundred gigabytes of model memory are available.

GLM-5.3-FlashOther modelsMedian

PriceInput / output per 1M tokens
$0.15 / $0.50Cheaper than 36 of 43 others
Context windowTokens in one request
1,000,000Larger than 7 of 46 others
Output speedTokens per second, our measurement
89 tokens/sFaster than 13 of 50 others
Developer
Z.AI
Released
26 Aug 2026
Family
GLM-5
License
MIT
Open model
Yes
Local model
Yes
Parameters
320B
Active parameters
18B
Max output
131,000 tokens
Input → output
Text, image → text
Architecture
Hybrid sparse and linear attention MoE

Run GLM-5.3-Flash locally

Hardware
Multi-GPU workstation/serverEstimated
Download size
200 GB (~4-bit)

Official Hugging Face weightsUnsloth GGUF

All prices and providers2

Vercel AI Gateway / DigitalOcean route; input 0.15 USD/1M; cached read 0.03/1M; output 0.5/1M; A specific gateway provider route; Vercel lists other providers with different prices and limits.

Local deployment details
  • Unsloth UD-Q4_K_XL GGUF200 GB · ~4-bit
Measurements and sources6

Our speed measurement: 88.98 tokens/s on average, from 18.65 to 127.37 tokens/s.

How StealthMark scores models · General and Visual Intelligence

GLM-5 family

  1. GLM-5.3-Flash48.9This model
  2. GLM-5.345.4Compare GLM-5.3-Flash and GLM-5.3

StealthMark score

StealthMark measures how well a model’s abilities carry over to original, demanding tasks it could not have trained for: physical and spatial reasoning, planning across many steps, and precise visual work. How StealthMark works explains every benchmark, what it tests and why it is hard.

General Intelligence

The ability to reason, plan, solve problems and understand how things relate and behave. A high score means it thinks hard problems through, from careful analysis to plans that span many steps.

Visual Intelligence

The ability to understand visual intent and turn ideas, descriptions and plans into coherent, effective visuals. A high score means design ideas come out the way you meant them, in illustrations, icons, diagrams or interface visuals.

Overall score

Overall shows how well a model’s abilities carry over to original, demanding tasks, weighting every benchmark into one score from 0 to 100. Estimated means the total includes an estimate for one missing scene.

Judging

Scores reflect how well each result meets the benchmark requirements. Game scores come from the recorded play: the points earned on each level and how each level ended. How StealthMark keeps scores honest.

Games

Each game is a sequence of levels, and earlier moves carry into later levels. A level that was not played is shown as not played and never counts as zero. Play every level of Latch or Foresight yourself to get a score calculated like the models’.

Reasoning efficiency

Good results with little thinking score high. On each task, a model’s result and reasoning tokens are compared with every other model and combined with the Balanced Integration Score (Liesefeld & Janczyk, 2019); 50 is average. Single step shows how much reasoning a model needs to understand and solve one problem, multi-step how efficiently it keeps working on a task by building on its earlier reasoning, as in agentic workflows. Not part of StealthMark.

BENCHMARK RESULTS

Compare results

This link opens the same results in the same order.

StealthMark benchmark results

Reset benchmark preferences?

Return to the default view, clear comparison selections and restore the advertisement.