GLM-5.3-Flash is a native multimodal model from Z.ai. It is suited for efficient coding and long-horizon agent tasks. Its hybrid sparse and linear attention architecture maintains accurate long-context behavior while...
Tested on 23 benchmarks · BenchGecko score 50.5. Top scores: Chatbot Arena Elo — Coding (1615.6%), Chatbot Arena Elo — Overall (1473.0%), OTIS Mock AIME 2024-2025 (93.9%).
Abstraction and Reasoning Corpus. Tests fluid intelligence through novel visual pattern recognition puzzles. Core measure of general intelligence.
ARC-AGI 2, harder sequel to ARC. More complex abstract reasoning patterns that test generalization ability beyond training data.
Mock AIME (American Invitational Mathematics Exam) problems from OTIS. Tests mathematical competition performance.
- Typemultimodal
- Context1.0M tokens (~524 books)
- ReleasedAug 2026
- LicenseProprietary
- StatusActive
- Cost / Message~$0.001
Frequently Asked Questions
Related Models
Gemini 2.0 Pro · Google DeepMindFalcon-180B · TIIPhi 3 Mini 4k Instruct · MicrosoftGemma 2 27B · Google DeepMindMagnum v4 72B · anthracite-orgKey facts · as of 2026-10-16
- GLM 5.3 Flash by z-ai. BenchGecko score 50.5, rank 150 of 319 scored models (normalized average of public benchmark scores).
- List price $0.15 input · $0.50 output per 1M tokens (as of 2026-10-16).
- Sold by 34 providers (as of 2026-10-16): OpenInference (fp4) $0.0211 in / $0.90 out · Relace $0.0212 in / $0.73 out · Morph (fp8) $0.0300 in / $1.00 out · StreamLake (fp8) $0.0435 in / $0.14 out · Sail Research (fp4) $0.0450 in / $0.60 out · and 29 more. Every provider
- Gecko Tests · Gecko Score 37 (rank 16): Who Are You B (Knows who made it) · World Map D (89.7% of the map right) · Tokenizer Tax D (70% more tokens outside English).
How to cite · data as of 2026-10-16
GLM 5.3 Flash · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-16. https://benchgecko.ai/model/glm-5-3-flash
Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP