Skip to content
ModelCap

GLM 5.3

Arena #29Official publisher releaseRestricted license6.9d #20 → #19

Z.ai·#19 of 371 on the current board·cheaper than #18 GPT-5.5

GLM 5.3 is a language model from Z.ai. It ranks #19 of 371 on ModelCap's current board, with an Index score of 78.5, measured on 3 public benchmarks. Its score range of 73.8–83.2 spans positions #7–#33. It is listed at $0.07 per million input tokens and $1.40 per million output tokens across 37 providers. It reads up to 1M tokens of context, and its weights are downloadable under a use-restricted licence.

Figures as of 19 Nov 2026, 05:16 UTC · download the public dataset · methodology

ModelCap Index · Current board
78.5#19
Measured on 3 public benchmarks · range 73.8–83.2
Evidence as of 18 Nov 2026, 00:00 UTC
Market Gravity
67.5Decreased by 1.7
Secondary market signal
Input
$0.07
per 1M tokens
Output
$1.40
per 1M tokens
Context
1M
tokens
Providers
37
49 endpoints
Gateway spend
10.47%
latest · Vercel share

Providers

Uptime measured over the last 30 minutes
ProviderInputOutputContextQuantUptime
FeatherlessCheapest$0.42$1.32262K—100.0%
Waferus$0.07$1.401Mfp4100.0%
Waferfp4$0.179$1.401Mfp4100.0%
Decartfp4$0.56$1.811Mfp4100.0%
SiliconFlowfp8$0.70$1.901Mfp8100.0%
Novitafp8$0.63$1.981Mfp8100.0%
DeepInfrafp4/flex$0.45$2.001Mfp4100.0%
Cloudflare$0.70$2.201M—100.0%
GMICloudfp8$0.70$2.201Mfp8100.0%
DeepInfrafp4$0.563$2.501Mfp4100.0%
Tenstorrentfp8$0.60$2.501Mfp8100.0%
DigitalOcean$0.91$2.861M—100.0%
CoreWeavenvfp4$0.75$3.001Mnvfp4100.0%
Sail Researchus$0.75$3.001Mfp8100.0%
Fireworksspot$0.98$3.081M—100.0%
Coherenvfp4$0.50$3.501Mnvfp4100.0%
InferenceNetus$0.084$3.601Mfp4100.0%
InferenceNetfp4$0.084$3.601Mfp4100.0%
Alibaba$1.19$3.741M—100.0%
Morphfp8$0.839$3.911Mfp8100.0%
Friendli$1.26$3.961M—100.0%
Inceptronfp4$1.05$4.081Mfp4100.0%
Sail Researchfp8$1.05$4.201Mfp8100.0%
Relacefp4$0.07$4.381Mfp4100.0%
AkashMLfp4$0.19$4.401Mfp4100.0%
AtlasCloudfp8$1.40$4.401Mfp8—
Baidufp8$1.40$4.401Mfp8—
BaseTenfp4/us$1.40$4.401Mfp4100.0%
BaseTenfp4$1.40$4.401Mfp4100.0%
Crusoefp4$1.40$4.401Mfp4—
Doubleword$1.40$4.401M——
Fireworks$1.40$4.401M—100.0%
Makorafp4$1.50$4.401Mfp4100.0%
Mistralzdr$1.40$4.401Mnvfp4100.0%
Modal$1.40$4.401M—100.0%
Nebiusfp4$1.40$4.401Mfp4100.0%
Parasailfp4$1.40$4.401Mfp4100.0%
SambaNova$1.40$4.401M——
Together$1.40$4.401M—100.0%
Venice$1.40$4.401M—100.0%
Z.AIfp8$1.40$4.401Mfp8100.0%
Mistralnvfp4/eu$1.54$4.841Mnvfp4100.0%
Alibabaprime$1.68$5.281M—100.0%
Amazon Bedrockus-east-1$1.68$5.281M—100.0%
BaseTenus/fast$2.10$6.601Mfp8100.0%
BaseTenfast$2.10$6.601Mfp8100.0%
Fireworksfast$2.10$6.601M—100.0%
ModelRunfp4$2.10$6.601Mfp4100.0%
Parasailfast$2.10$6.601Mfp4100.0%

Hugging Face Inference Providers

Exact repository identity

Provider economics and telemetry published by the Hugging Face Router. These are operational signals, not benchmark results.

ProviderInputOutputContextTTFTThroughputFeatures
featherless-ai
live
$0.42$1.32262K1086 ms66.0 tok/stools
wafer
live
$0.07$1.401M322 ms90.0 tok/stools · structured
decart
live
$0.56$1.811M570 ms102.0 tok/stools · structured
siliconflow
live
$0.70$1.901M1836 ms41.0 tok/stools
novita
live
$0.63$1.981M2179 ms46.0 tok/stools · structured
cloudflare
live
$0.70$2.201M3521 ms42.0 tok/stools · structured
deepinfra
live
$0.563$2.501M938 ms65.0 tok/stools · structured
tenstorrent
live
$0.60$2.501M710 ms114.0 tok/stools
digitalocean
live
$0.91$2.861M505 ms97.0 tok/stools · structured
coreweave
live
$0.75$3.001M425 ms101.0 tok/stools · structured
sail-research
live
$0.75$3.001M501 ms64.0 tok/stools · structured
fireworks-ai
live
$0.98$3.081M619 ms68.0 tok/stools · structured
cohere
live
$0.50$3.501M739 ms84.0 tok/stools · structured
inference-net
live
$0.084$3.601M306 ms95.0 tok/stools · structured
morph
live
$0.839$3.921M786 ms88.0 tok/stools · structured
inceptron
live
$1.05$4.081M945 ms96.0 tok/stools
relace
live
$0.07$4.381M1078 ms74.0 tok/stools
akashml
live
$0.19$4.401M1118 ms92.5 tok/stools · structured
baseten
live
$1.40$4.401M1340 ms72.0 tok/stools · structured
crusoe
live
$1.40$4.401M577 ms62.0 tok/stools · structured
doubleword
live
$1.40$4.401M795 ms56.0 tok/stools · structured
makora
live
$1.50$4.401M250 ms119.0 tok/stools · structured
mistral
live
$1.40$4.401M590 ms97.0 tok/stools · structured
modal
live
$1.40$4.401M315 ms101.0 tok/stools · structured
nebius
live
$1.40$4.401M1466 ms108.0 tok/stools · structured
nscale
live
$1.40$4.401M485 ms116.4 tok/stools · structured
parasail
live
$1.40$4.401M260 ms123.0 tok/stools · structured
sambanova
live
$1.40$4.401M1008 ms47.0 tok/stools · structured
together
live
$1.40$4.401M556 ms97.0 tok/stools · structured
venice
live
$1.40$4.401M596 ms103.0 tok/stools · structured
zai-org
live · model author
$1.40$4.401M2096 ms57.0 tok/stools · structured
amazon-bedrock
live
$1.68$5.281M2121 ms83.0 tok/stools · structured
modelrun
live
$2.10$6.601M450 ms114.5 tok/stools · structured

Reasoning efforts

Best supported result can inform ModelCap

Published Max, X-High, High and other configurations for this model. These are sourced scores, not head-to-head essays.

EffortAgent scoreSource rankSampleSource
Max
GLM-5.3 (Max)
-0.04%
CI -0.9–0.8%
#3611.4M obs
95K sessions
LMArena Agent
Match 100%

Related community configurations

Human preference rating
ConfigurationRatingSource rankVotesSource
Max
GLM-5.3 (Max)
1483
CI 1478–1487
#2934KLMArena
Match 100%

Reported evaluation leads (11)

Not independent measurement

Structured rows reported in this exact model repository at the pinned revision below. They are the publisher talking about itself: never measured evidence, never a score input on their own. They remain quarantined until dataset revision, harness, configuration and identity pass ModelCap’s independent admission review.

Dataset / taskMetricValueRevisionProvenance
agents-last-exam/agents-last-exam
ale-cli-pass-rate
—28.5
Unpinned
not published
Reported
Quarantined
cais/hle
hle
—62.5
Unpinned
not published
Reported
Quarantined
collinear-ai/cwe-bench
resolve_rate
—44
Unpinned
not published
Reported
Quarantined
datacurve/deep-swe
deep_swe
—66.9
Unpinned
not published
Reported
Quarantined
harborframework/terminal-bench
terminalbench_3
—28.3
Unpinned
not published
Reported
Quarantined
harborframework/terminal-bench
terminalbench_4_1
—41.2
Unpinned
not published
Reported
Quarantined
harborframework/terminal-bench
terminalbench_4
—41.9
Unpinned
not published
Reported
Quarantined
harborframework/terminal-bench-2.1
terminalbench_2_1
—88.2
Unpinned
not published
Reported
Quarantined
harborframework/terminal-bench-science
terminalbenchscience_0_1
—9.5
Unpinned
not published
Reported
Quarantined
hkust-nlp/Toolathlon
toolathlon_verified
—73
Unpinned
not published
Reported
Quarantined
IntelligenceLab/Long-Horizon-Terminal-Bench
lhtb2_solved
—12
Unpinned
not published
Reported
Quarantined

Other versions (6)

ModelCap Index · Current board

Methodology
78.5
Measured on 3 public benchmarks
Current board
#19 of 371
Score range
73.8–83.2
Rank range
#7–#33
Evidence as of
18 Nov 2026, 00:00 UTC
Benchmarks used
Arena Coding, Arena Overall, LMArena Agent
Technical details
Public rank basis
Measured (measured)
Evidence label
Measured · 3 public benchmark observations across 3 boards
Index support
65.7% · strong
Identity binding
Exact catalogue product · z-ai/z-ai/glm-5.3/fp8 · aggregates disclosed benchmark configurations · endpoint configuration not claimed · artifact metadata zai-org/GLM-5.3@ba32f8c19c8738e2c57c71798c601d477a7bf844 (not the evaluation revision)
Index observations used
3
Qualified Index evidence point
78.5
Bayesian Evidence Score · measured evidence input only

BES normalizes admitted public benchmark evidence for measured Index rows. Its legacy score and observables prior do not define the public language rank.

Observed capability
78.5
Evidence status
confirmed · 88% mass
General preference
79.9
Coding
78.3
Agents & tools
51.1
Reasoning
—
Evidence breadth
97%
Evidence coverage
66%
Market and catalogue signals · secondary, never the language Index rank
Usage (OpenRouter popularity)
57.5
Liquidity (providers × uptime)
99.6
Open reach (HF downloads)
—
Surface (context / tools / modalities)
85.0
Freshness
70.3
  • · OpenRouter popularity #12
  • · 37 live providers
  • · Catalogue: context 1048576, reasoning, tools
  • · First seen 2026-08-18

These adoption and deployment observations remain context only. They do not change this model's ModelCap Index score or rank.

BES evidence selected · measured input only

The bounded 0–100 capability score blends each source’s competitive placement with its published achievement, then combines capability families. Evidence breadth adds a modest uncertainty adjustment; price and popularity are not part of benchmark-led rank.

Deployment readiness

Metadata index
84.3
strong · 89% metadata coverage
Not capability
Artifact reproducibility
85.0
Access & legal clarity
67.5
Deployability
100.0
Evaluation provenance
76.6

Not yet published: the base model it derives from, a link to the licence terms, pinned evaluation dataset versions.

A metadata completeness and deployability index—not a safety certification, quality grade, or production-readiness claim.

Market activity

Venue coverage
Market Gravity
67.5
OpenRouter weekly popularity
#12 of 345
Vercel token share · 7d
9.11%
Vercel request share · 7d
6.29%
Venue data through
2026-11-19

Market Gravity uses OpenRouter’s full-catalogue popularity order. Sparse Vercel top-ten observations appear only when published and do not affect the score.

Arena preference

Status
Ranked
Source
LMArena
Official rank
#29 of 422
Rating
1482.6
95% confidence interval
1478–1487
Votes
34K
Source published at
18 Nov 2026, 00:00 UTC
Category
overall
Identity match confidence
100%

Market Gravity breakdown

Usage55%
57.5

OpenRouter weekly popularity across the full model catalogue.

Liquidity25%
99.6

Independent providers versus the model's open or closed cohort, weighted by uptime.

Open reach15%
50.0

Hugging Face 30-day downloads for open models; neutral for closed models.

Freshness5%
70.3

Time since first public availability, on a six-month half-life.

Specification

Model ID
z-ai/glm-5.3
Context window
1M tokens
Declared output cap
944K tokens

OpenRouter catalogue or provider-endpoint declaration. Limits can vary by provider and prompt length; a model-wide maximum has not been independently verified. OpenRouter metadata

Inputs
Text
Outputs
Text
Tokenizer
Other
Cached input tokens
$0.069 / 1M
First seen on OpenRouter
18 Aug 2026

Capabilities

  • Supported: Reasoning
  • Supported: Tool use
  • Supported: Structured output
  • Supported: Response format
  • Not supported: Moderated

Weights & access

Restricted license. Downloadable weights use a model-specific or use-restricted license. Review the linked license before use, redistribution, or commercial deployment.

Access
Restricted license
Licence
glm-5.3
Pinned revision
ba32f8c19c87
Parameters
753.3B
Architecture
GlmMoeDsaForCausalLM
Model type
glm_moe_dsa
Hugging Face downloads (30d)
1.7M
Downloads (all time)
4.5M
Likes
2K
Repository
zai-org/GLM-5.3
Repository updated
14 Oct 2026, 12:22 UTC

Compare GLM 5.3 with

Pick any model

Benchmark leaderboards featuring GLM 5.3

All leaderboards
Public benchmark boards on which GLM 5.3 has a published result
LeaderboardScoreSource rankConfigurationPublished
Arena text leaderboardArena (LMArena)14831478–1487#29 of 422Max18 Nov 2026
Arena coding leaderboardArena (LMArena)15231516–1531#33 of 417Max18 Nov 2026
LMArena Agent leaderboardLMArena0.0%-0.9%–0.8%#36 of 68Max18 Nov 2026
Artificial Analysis Intelligence IndexArtificial Analysis31.7#53 of 125Max28 Oct 2026
Terminal-Bench 4.0 leaderboardTerminal-Bench (Laude Institute)41.8%38.6%–45.1%#20 of 35Claude Code · Max14 Aug 2026

Alternatives to GLM 5.3, by the numbers

Ranked neighbours

Models within 3 places of #19 on the ModelCap Index.

Cheaper at a similar or higher Index

Index score within 3 points of, or above, this model's, with a lower listed blended price (3:1 input:output).

Other ranked Z.ai models

The same publisher's other current models with a public Index position.

GLM 5.3: common questions

How does GLM 5.3 rank among AI models?

GLM 5.3 holds ModelCap Index position #19 of 371 ranked language models as of 19 November 2026, with an Index score of 78.5 (interval 73.8–83.2); evidence: measured. The Index combines public benchmark boards with published uncertainty; the methodology page explains the weighting.

How much does GLM 5.3 cost per 1M tokens?

GLM 5.3 is listed at $0.07 per 1M input tokens and $1.40 per 1M output tokens as of 19 November 2026; the lowest output price among 37 providers is $1.32 on Featherless. Prices come from the OpenRouter catalogue and re-observe on every refresh. For 1,000 requests using 2,000 input and 500 output tokens each, the listed rates imply $0.84 total (2M input + 0.5M output). This excludes caching, batch discounts, prompt-length tiers, tool charges and retries; verify the chosen endpoint's rate and limits before budgeting.

What is the context window of GLM 5.3?

GLM 5.3 has a published context window of 1M tokens (1,048,576). The snapshot lists a 943,718 token catalogue-declared route output cap. OpenRouter catalogue or provider-endpoint declaration. Limits can vary by provider and prompt length; a model-wide maximum has not been independently verified. The providers table lists each endpoint's context limit.

Is GLM 5.3 open-weight?

Restricted license (glm-5.3). Downloadable weights use a model-specific or use-restricted license. Review the linked license before use, redistribution, or commercial deployment. The repository is zai-org/GLM-5.3 on Hugging Face.

Which API providers serve GLM 5.3?

37 providers serve GLM 5.3 through OpenRouter as of 19 November 2026: Wafer, Relace, InferenceNet, AkashML, Featherless and DeepInfra and 31 more. Each provider's price, context limit, quantization and measured uptime are in the providers table above.

Which benchmarks has GLM 5.3 been evaluated on?

GLM 5.3 has published results on 5 public boards tracked by ModelCap as of 19 November 2026: Arena text 1483 (#29 of 422), Arena coding 1523 (#33 of 417), LMArena Agent 0.0% (#36 of 68), AA Intelligence Index 31.7 (#53 of 125) and Terminal-Bench 4.0 41.8% (#20 of 35). Each board page ranks every tracked model on that benchmark. The ModelCap Index scores the results its methodology admits, with published uncertainty, and shows reference boards without scoring them.

How does GLM 5.3 compare with GPT-5.5?

GLM 5.3 ranks #19 (Index 78.5) and GPT-5.5 ranks #18 (Index 78.8) on the ModelCap Index as of 19 November 2026. The head-to-head page lines up their benchmarks, listed prices, context windows and provider counts side by side.

When was GLM 5.3 released?

GLM 5.3 first appeared in the catalogue on 18 Aug 2026. It is the current version in its family. Rank and price movements since then are recorded on the site's changes feed.