The Known Good Updated 25 Jul 2026
DeepSeek: DeepSeek V4 Pro vs Z.ai: GLM 5.2 ·DeepSeek vs Z.ai

DeepSeek: DeepSeek V4 Pro vs Z.ai: GLM 5.2 — head-to-head comparison

DeepSeek: DeepSeek V4 Pro profile Z.ai: GLM 5.2 profile All comparisons
DeepSeek: DeepSeek V4 Pro — DeepSeek
Quality
#6 of 61
93based on 3 of 6 evaluations
Known Good Index
Speed
#182 of 316
45.2
Output tokens per second
Price
#145 of 340
$0.54
Blended per 1M tokens
In $0.43 · Out $0.87
Cache hit price
#12 of 206
$0.004
per 1M tokens (−99%)
Context
#25 of 345
1.05M
tokens (~1,600 pages)
Z.ai: GLM 5.2 — Z.ai
Quality
#8 of 61
91based on 3 of 6 evaluations
Known Good Index
Speed
#147 of 316
55.2
Output tokens per second
Price
#204 of 340
$1.07
Blended per 1M tokens
In $0.70 · Out $2.20
Cache hit price
#115 of 206
$0.13
per 1M tokens (−81%)
Context
#45 of 345
1.05M
tokens (~1,600 pages)
Summary

On the Known Good Index, DeepSeek: DeepSeek V4 Pro scores 93 and Z.ai: GLM 5.2 scores 91 — a 2-point lead for DeepSeek: DeepSeek V4 Pro. They share 3 published evaluations, with DeepSeek: DeepSeek V4 Pro ahead on 1 and Z.ai: GLM 5.2 ahead on 2.

Blended pricing is $0.54 per 1M tokens for DeepSeek: DeepSeek V4 Pro and $1.07 for Z.ai: GLM 5.2 — DeepSeek: DeepSeek V4 Pro is about 2.0× cheaper.

Throughput is 45 tokens per second for DeepSeek: DeepSeek V4 Pro and 55 for Z.ai: GLM 5.2, with time to first token at 2.0s and 1.7s. DeepSeek: DeepSeek V4 Pro accepts 1.05M tokens of context against 1.05M.

DeepSeek: DeepSeek V4 Pro is open weights under Open weights (unrestricted); Z.ai: GLM 5.2 is open weights under Open weights (unrestricted). 30 tracked hosts serve one or both, and 17 serve both.

Generated from tracked data on 25 Jul 2026 · how these comparisons are computed →

At a glance
MetricDeepSeek: DeepSeek V4 ProZ.ai: GLM 5.2
Known Good Index 93 based on 3 of 6 evaluations 91 based on 3 of 6 evaluations
Blended $/1M $0.54 $1.07
Output tok/s 45 55
Context window 1.05M 1.05M
Providers 18 29
Jump to section

Quality

Known Good Index and the capability indices, side by side

IndicesComponents
Index scores

Each index is min-max normalised across tracked models and scaled 0–100 from published evaluations. We do not run these evaluations.

MetricDeepSeek: DeepSeek V4 ProZ.ai: GLM 5.2
Known Good Index 93 91
Mathematics Index 97 86
Science Index 94 97
Agentic Index 89 91
Openness Index 60 60
The Known Good

Quality Breakdown

Every evaluation both models report a published score for

Evaluations both models report

Bar = DeepSeek: DeepSeek V4 Pro · marker = Z.ai: GLM 5.2 · scores ingested from Epoch AI (CC-BY 4.0)

GPQA Diamond90 / 92
MATH / AIME97 / 86
SWE-bench Verified78 / 79

Right-hand figures read DeepSeek: DeepSeek V4 Pro / Z.ai: GLM 5.2. DeepSeek: DeepSeek V4 Pro leads on 1 of 3, Z.ai: GLM 5.2 on 2.

The Known Good

Arena Elo

Human preference rating from blind pairwise votes

Arena ratings

Elo from blind pairwise votes, ingested from Arena. Higher is better.

MetricDeepSeek: DeepSeek V4 ProZ.ai: GLM 5.2
Text Arena 1,449 1,464
Text Arena (style controlled) 1,457 1,469
The Known Good

Price & Cost

Published API pricing by token type

Per token typeCache discount
Published pricing

USD per 1M tokens. Blended is a 3:1 input:output mix. Live from OpenRouter, cross-checked against models.dev. Lower is better.

MetricDeepSeek: DeepSeek V4 ProZ.ai: GLM 5.2
Input $/1M $0.43 $0.70
Output $/1M $0.87 $2.20
Blended $/1M $0.54 $1.07
Cache read $/1M $0.004 $0.130
The Known Good
We publish blended price rather than cost-per-task. Cost-per-task requires token counts from an evaluation suite we do not run. Blended price is computed from published rates and is fully reproducible — see the pricing methodology.

Context Window

Maximum input accepted and maximum output emitted

Context and output limits

Provider-declared limits, from metadata via models.dev and OpenRouter. Higher is better.

MetricDeepSeek: DeepSeek V4 ProZ.ai: GLM 5.2
Context window 1.05M 1.05M
Max output tokens 384k 131k
The Known Good

Speed & Latency

Median throughput and time to first token across hosting providers

Output speedTime to first token
Throughput and latency

Medians across all tracked providers, derived from OpenRouter endpoint telemetry.

MetricDeepSeek: DeepSeek V4 ProZ.ai: GLM 5.2
Output tokens/s 45 55
Time to first token 2.0s 1.7s
The Known Good

Providers

Which hosts serve each model, and at what price and speed

30 hosts · 17 serve both
Full provider benchmarks →
Provider DeepSeek: DeepSeek V4 Pro blendedDeepSeek: DeepSeek V4 Pro tok/s Z.ai: GLM 5.2 blendedZ.ai: GLM 5.2 tok/s
BaseTen $2.17 110 $2.15 120
Fireworks $2.17 66 $2.15 123
Novita $1.46 66 $1.07 39
DeepInfra $1.62 65 $1.45 68
GMICloud $0.85 62 $1.42 50
Venice $2.06 61 $2.15 56
Baidu · first party $0.78 58 $1.12 42
Together $2.17 53 $2.15 62
Alibaba · first party $1.77 52 $1.27 65
AtlasCloud $2.10 46 $1.94 34
Wafer $1.50 45 $2.15 108
DeepSeek · first party $0.54 44 — —
SiliconFlow $1.91 41 $2.00 49
StreamLake $0.84 41 $1.07 45
CoreWeave $2.17 28 $1.18 49
Ionstream $1.41 20 $2.15 108
DigitalOcean $1.74 6 $1.89 23
Parasail $2.17 — $2.15 45
Friendli — — $2.15 124
Cloudflare — — $2.15 84
Ambient — — $1.89 78
Decart — — $1.52 71
Chutes — — $1.93 57
Morph — — $1.85 52
Z.AI · first party — — $2.15 49
Phala — — $2.15 48
Inceptron — — $1.43 34
AkashML — — $1.18 33
Io Net — — $4.40 28
Sail Research — — $1.72 27

Per-host pricing and telemetry from OpenRouter. A dash means that host does not serve that model.

The Known Good
Sources for this page
Pricing & providers — OpenRouter Metadata — models.dev Benchmarks — Epoch AI (CC-BY 4.0)Elo — Arena Full attribution →