Z.AI · Model release
Data as of October 10, 2026 · How the score is built
GLM-5.3-Flash
Decision readingGLM-5.3-Flash scores 57.4 out of 100 and ranks #55 of 218. This profile shows 21 source-displayable benchmark rows; its strongest eligible category is Multimodal & Grounded at #14. Published weights can be self-hosted, but infrastructure cost varies and is not a comparable API token rate.
Released Aug 26, 2026 — see all recent releases
GLM-5.3-Flash will be repriced, updated or retired. Get each notice with its source and date. Follow model changes
Decision snapshot
Each value carries a field reference instead of floating alone. Markers compare this model with the current ranked and priced catalog; they are not absolute quality thresholds.
Capability
57.4/100
field median 50.1#55 of 218 ranked models
Public
#55of 218
Verified —
Price
Self-hosted; infrastructure cost varies
input median $0.95No comparable first-party hosted token rate
Speed
Not measured
field median 97 tok/sTime to first token not measured
Context
1Mtokens
field median 256,000Maximum output length is tracked separately
Strongest published evidence
Multimodal & Grounded ranks #14. Particularly strong for screenshots, documents, charts, and grounded multimodal workflows.
Validate before choosing
21 published rows leave some tracked benchmark slots empty. No comparable first-party API token rate is published.
Source-linked · 21 displayable benchmark rows
Follow model changesCategory score record
Scores and ranks appear only where published evidence can be displayed. The table keeps the score, weight, cohort, and evidence state together.
| Category | Score | Rank | Percentile | Weight | Benchmarks | Evidence |
|---|---|---|---|---|---|---|
| AgenticRank #34 of 123Percentile 73rdWeight 22%6 benchmarksVerified | 55.7 | 6 benchmarks | Verified | |||
| CodingRank #42 of 146Percentile 72ndWeight 20%8 benchmarksVerified | 49.3 | 8 benchmarks | Verified | |||
| ReasoningWeight 17%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MultimodalRank #14 of 54Percentile 75thWeight 12%5 benchmarksVerified | 84.0 | 5 benchmarks | Verified | |||
| KnowledgeRank #43 of 177Percentile 76thWeight 12%2 benchmarksVerified | 60.5 | 2 benchmarks | Verified | |||
| MultilingualWeight 7%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| Inst. FollowingWeight 5%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured | |||
| MathWeight 5%0 benchmarksNot measured | Not measured | 0 benchmarks | Not measured |
21 of 667 tracked benchmark slots have displayable evidence · bars run 0–100
Coverage detailsHow much of this is verified
Coverage is split by category so a strong number never hides a thin evidence base. Verified means the row is tied to a published source; provisional rows remain visible but separate.
- Agentic6/6 verified
- Coding8/8 verified
- ReasoningNot measured
- Multimodal5/5 verified
- Knowledge2/2 verified
- MultilingualNot measured
- Inst. FollowingNot measured
- MathNot measured
Capability shape
Each axis shows percentile within that category’s eligible cohort. The comparison outline is the median of the six nearest public-score peers; a collapsed vertex means the category is not rank-eligible.
GLM-5.3-Flash category percentile values
- Agentic73rd percentile
- Coding72nd percentile
- ReasoningNot eligible
- Multimodal75th percentile
- Knowledge76th percentile
- MultilingualNot eligible
- Instruction followingNot eligible
- MathNot eligible
The dashed outline is median of 6 nearest peers.
Eligible category ranks
- Agentic#34/123
- Coding#42/146
- ReasoningNot ranked
- Multimodal#14/54
- Knowledge#43/177
- MultilingualNot ranked
- Inst. FollowingNot ranked
- MathNot ranked
Benchmark ledger
Coding opens by default. The marker compares each value with the best source-verified result in the catalog; provisional leaders do not set the reference. Expand the remaining categories for every published row.
Coding8 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| DeepSWE | Score63.4% | Versus best verified row Best verified: Gemini 4 Argon · 77.9% | Gap14.5 behind | Weight15% ref. weight | Provider exact |
| LiveCodeBench (Vals)LiveCodeBench, Vals AI run | Score80.5% | Versus best verified row Best verified: Claude Fable 5.1 · 90.5% | Gap10 behind | Weight8% ref. weight | |
| FrontierSWE v2 | Score18.1% | Versus best verified row Best verified: GPT-6 Astra · 65.5% | Gap47.4 behind | Weight8% ref. weight | Benchmark exact |
| Terminal-Bench 2.1 | Score84.3% | Versus best verified row Best verified: SWE-2 · 92.8% | Gap8.5 behind | WeightScored in Agentic | Provider exact |
| NL2Repo | Score56.3% | Versus best verified row Best verified: DeepSeek V4.1 Flash · 65.4% | Gap9.1 behind | WeightDisplay only | Provider exact |
| SWE-bench (Vals)SWE-bench, Vals AI run | Score92.0% | Versus best verified row Best verified: Claude Opus 5 · 97.0% | Gap5 behind | WeightDisplay only | Verified |
| OpenHarmony BenchOpenHarmony Bench v1.0 | Score57.3% | Versus best verified row Best verified: Qwen3.8 Max · 60.8% | Gap3.5 behind | WeightDisplay only | Benchmark exact |
| Bug Hunt Bench | Score17.7 fixes | Versus best verified row Best verified: Claude Sonnet 5.5 · 51.3 fixes | Gap33.6 behind | WeightDisplay only | Benchmark exact |
Agentic6 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| Terminal-Bench 2.1 | Score84.3% | Versus best verified row Best verified: SWE-2 · 92.8% | Gap8.5 behind | Weight8% ref. weight | Provider exact |
| AutomationBench | Score48.8% | Versus best verified row Best verified: DeepSeek V4.1 Flash · 54.8% | Gap6 behind | Weight5% ref. weight | Provider exact |
| Terminal-Bench 2.1 (Vals)Terminal-Bench 2.1, Vals AI run | Score62.9% | Versus best verified row Best verified: GPT-6 Astra · 87.3% | Gap24.4 behind | Weight3% ref. weight | |
| Toolathlon-Verified | Score78.4% | Versus best verified row Best verified: Claude Opus 5 · 80.6% | Gap2.2 behind | Weight3% ref. weight | Provider exact |
| Agents' Last Exam | Score26.3% | Versus best verified row Best verified: GPT-6 Astra · 59.3% | Gap33 behind | Weight3% ref. weight | Provider exact |
| HLE w/ toolsHumanity's Last Exam with tools | Score55.3% | Versus best verified row Best verified: Claude Opus 5.5 · 67.7% | Gap12.4 behind | Weight3% ref. weight | Provider exact |
Multimodal5 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| OfficeQA Pro | Score62.4% | Versus best verified row Best verified: Claude Opus 5.5 · 67.7% | Gap5.3 behind | WeightWeighted 25% | Provider exact |
| CharXivCharXiv Reasoning | Score89.4% | Versus best verified row Best verified: Qwen3.8 Max · 93.5% | Gap4.1 behind | WeightWeighted 20% | Provider exact |
| Chartography (tools)Chartography with image and code tools | Score78.0% | Versus best verified row Best verified: Claude Sonnet 5.5 · 90.2% | Gap12.2 behind | WeightDisplay only | Provider exact |
| BabyVision | Score53.4% | Versus best verified row Best verified: Qwen3.8 Max · 82.0% | Gap28.6 behind | WeightDisplay only | Provider exact |
| MMVUMultimodal Multi-disciplinary Video Understanding | Score80.5% | Versus best verified row Best verified: Qwen3.8 Max · 82.4% | Gap1.9 behind | WeightDisplay only | Provider exact |
Knowledge2 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| MMLU-Pro (Vals)MMLU-Pro, Vals AI run | Score86.1% | Versus best verified row Best verified: Claude Fable 5.1 · 92.4% | Gap6.3 behind | Weight6% ref. weight | Verified |
| GPQA Diamond (Vals)GPQA Diamond, Vals AI run | Score86.4% | Versus best verified row Best verified: Gemini 3.1 Pro · 95.5% | Gap9.1 behind | Weight2% ref. weight |
Lineage
The sequence follows explicit supersedes links. Each score is estimated for that model; a relative can inform a sparse estimate but never sets a floor, so a newer release can score below an earlier one. Scores and prices remain blank when the corresponding public row or first-party rate is unavailable.
GLM-5.3-Flash release history
Radar confirmed these at the source. Use GLM-5.3-Flash in your work? Explore Radar to follow supported changes and choose your alerts.
Radar
Spec sheet
Each documented value carries its source. Missing fields stay visible as not sourced or not published, rather than disappearing from the page.
- API model ID
- glm-5.3-flashGLM-5.3-Flash model card
- Context window
- 1MGLM-5.3-Flash model card
- Maximum output
- Not sourced yet
- Knowledge cutoff
- Not sourced yet
- Input modalities
- Not sourced yet
- Output modalities
- Not sourced yet
- Parameters
- Not sourced yet
- Availability
- Z.AI publishes the FP8 GLM-5.3-Flash checkpoint on Hugging Face under MIT and documents local serving with SGLang, vLLM, TokenSpeed, and KTransformers. The launch post says the exact glm-5.3-flash API model is available to GLM Coding Plan users and through Z.AI API services, but the public pricing table does not yet list a standalone per-token rate for this SKU.
- Cloud regions
- Not tracked yet
- Lifecycle
- Current
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Not documented in the pricing recordGLM-5.3-Flash model card
- Self-host
- Open weights available; hardware estimate not sourced
- Rate limits
- Not tracked yet
How to read this profile
The visual layer above carries the decisions. These notes preserve the model, ranking, coverage, and family context behind the numbers.
GLM-5.3-Flash ranks #55 of 218 on the public leaderboard with a score of 57.36/100. It does not yet have enough sourced coverage for a verified position.
GLM-5.3-Flash is a open weight model with a 1M context window. It uses an explicit reasoning mode, which can improve complex problem solving while adding latency and token use.
Z.AI publishes the FP8 GLM-5.3-Flash checkpoint on Hugging Face under MIT and documents local serving with SGLang, vLLM, TokenSpeed, and KTransformers. The launch post says the exact glm-5.3-flash API model is available to GLM Coding Plan users and through Z.AI API services, but the public pricing table does not yet list a standalone per-token rate for this SKU.
GLM-5.3-Flash sits in the GLM-5 family with GLM-5, GLM-5.2, GLM-5.1, GLM-5.3, GLM-5-Turbo, GLM-5V-Turbo, GLM-5 (Reasoning). Its explicit predecessor is GLM-4.7-Flash. 21 of 667 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Its strongest eligible category is Multimodal & Grounded at #14, while its lowest eligible position is Knowledge at #43. particularly strong for screenshots, documents, charts, and grounded multimodal workflows.
Last updated October 10, 2026. Runtime fields remain blank until a sourced snapshot exists.
Deployment options
Self-host and provider-specific paths stay separate from benchmark evidence so operating constraints are visible before a score becomes the whole decision.
Published weights are available, but BenchLM does not yet have a sourced parameter and VRAM profile for this exact model. Hardware cost estimates stay unavailable until that sizing record is complete.
Estimate VRAM from known parametersQuestions
How does GLM-5.3-Flash perform overall in AI benchmarks?
GLM-5.3-Flash ranks #55 out of 218 models on the public BenchAlign leaderboard, with a score of 57.36/100. Its evidence status is Estimated, and this profile shows 21 source-displayable benchmark rows. The label describes evidence depth, not a provider quality claim; inspect category rows before choosing a workload.
Is GLM-5.3-Flash good for knowledge and understanding?
GLM-5.3-Flash ranks #43 out of 177 eligible models for knowledge and understanding, with a public category score of 60.5/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is GLM-5.3-Flash good for coding and programming?
GLM-5.3-Flash ranks #42 out of 146 eligible models for coding and programming, with a public category score of 49.3/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is GLM-5.3-Flash good for agentic tool use and computer tasks?
GLM-5.3-Flash ranks #34 out of 123 eligible models for agentic tool use and computer tasks, with a public category score of 55.7/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is GLM-5.3-Flash good for multimodal and grounded tasks?
GLM-5.3-Flash ranks #14 out of 54 eligible models for multimodal and grounded tasks, with a public category score of 84/100. Higher-ranked alternatives are available for workloads where this category decides the choice. Check the underlying rows before treating the aggregate as a workload guarantee.
Is GLM-5.3-Flash open source?
GLM-5.3-Flash is an open-weight model from Z.AI. Its weights can be downloaded for local or hosted deployment, subject to the published license. Open weight does not automatically mean open source: training data and training code may remain private, and commercial restrictions can still apply.
Which sibling models are related to GLM-5.3-Flash?
GLM-5.3-Flash belongs to the GLM-5 family. Related tracked variants include GLM-5, GLM-5.2, GLM-5.1, plus 4 more. A sibling link indicates shared lineage or a documented configuration relationship; it does not mean the variants have identical pricing, context limits, benchmark evidence, or deployment behavior. Compare before switching.
Does GLM-5.3-Flash have full benchmark coverage on BenchLM?
No. GLM-5.3-Flash currently has 39 source-displayable rows across 667 tracked benchmark slots. The profile exposes published, non-generated evidence and leaves missing categories blank until an exact evaluation is available. Coverage describes how much was measured; it is not a penalty added to an individual benchmark result.
What is the context window size of GLM-5.3-Flash?
GLM-5.3-Flash has a documented context window of 1M. That figure is the maximum combined prompt and retained-conversation space reported for this exact model; it is not the maximum output length. The profile keeps output limits separate because providers often publish those limits independently.