vs
3:8
overall capability wins (0 ties across 11 scored). GLM-5.1 leads overall.
- Reasoning2:4
- Coding1:1
- Arena0:1
- Pricing3:0
- Specs0:2
Verdict
GLM-5.1 edges reasoning & knowledge; GLM-5.1 ranks higher on LMArena (+8 Elo); GLM-5.1 offers larger context (203K vs 200K).
- GLM-5.1 edges reasoning & knowledge
- GLM-5.1 ranks higher on LMArena (+8 Elo)
- GLM-5.1 offers larger context (203K vs 200K)
- GLM-5.1 allows ~3.9x more max output (128K vs 33K)
- GLM-5 is ~3.6x cheaper on a blended token basis than GLM-5.1
Cost for 1M in + 250K out
Illustrative chat workload at primary-provider list prices.
- GLM-5
- $0.70
- GLM-5.1
- $2.50
GLM-5 is about 3.6x cheaper on this workload.
Price per 1M tokens
Full comparison
Organization
Zhipu AI
Zhipu AI
Family
GLM
GLM
License
MIT
MIT
Open weights
Yes
Yes
Release
Feb 1, 2026
Apr 7, 2026
Knowledge cutoff
-
-
API / provider
Zhipu / Z.ai
Zhipu / Z.ai
Modalities
text → text
text → text
Specs
Context window
200K
203K
Max output
33K
128K
Parameters
—
744B (40B act.)
Pricing
Input $/1M
$0.40
$1.40
Output $/1M
$1.20
$4.40
Blended $/1M (3∶1)
$0.60
$2.15
Speed
tok/s
80
-
TTFT (s)
0.35
-
Reasoning
MMLU-Pro
—
—
GPQA Diamond
82.0
86.8
Humanity's Last Exam
29.3
30.1
AIME 2025
—
—
MATH-500
—
—
Humanity's Last Exam (with tools)
—
—
AA-Omniscience Accuracy
26.3
23.7
AA-LCR v1.1
75.7
73.7
CritPt
2.0
4.6
MMMU-Pro
—
—
IFBench
72.3
76.3
Chartography
—
—
Chartography (With Tools)
—
—
Coding
SWE-bench Verified
72.8
—
SWE-bench Pro
—
—
SWE-bench Multilingual
69.7
—
LiveCodeBench
—
—
Terminal-Bench 2.1
—
61.8
Aider Polyglot
—
—
Terminal-Bench 3
—
—
BigCodeBench
—
—
SciCode
46.2
44.8
CursorBench
—
—
SWE-Rebench
53.3
—
NL2Repo-Bench
—
—
DeepSWE
—
—
WebDev Arena
1,436
1,508
Terminal-Bench 4.0
—
—
CursorBench 4.0
—
—
FrontierCode v1.1 (Main)
—
—
Arena
LMArena Elo
1,458
1,466
| Metric | GLM-5 | GLM-5.1 | Delta |
|---|---|---|---|
| Identity | |||
| Organization | Zhipu AI | Zhipu AI | — |
| Family | GLM | GLM | — |
| License | MIT | MIT | — |
| Open weights | Yes | Yes | — |
| Release | Feb 1, 2026 | Apr 7, 2026 | — |
| Knowledge cutoff | - | - | — |
| API / provider | Zhipu / Z.ai | Zhipu / Z.ai | — |
| Modalities | text → text | text → text | — |
| Specs | |||
| Context window | 200K | 203K | B +3K |
| Max output | 33K | 128K | B +95K |
| Parameters | — | 744B (40B act.) | — |
| Pricing | |||
| Input $/1M | $0.40 | $1.40 | A +$1.00 |
| Output $/1M | $1.20 | $4.40 | A +$3.20 |
| Blended $/1M (3∶1) | $0.60 | $2.15 | A +$1.55 |
| Speed | |||
| tok/s | 80 | - | — |
| TTFT (s) | 0.35 | - | — |
| Reasoning | |||
| MMLU-Pro | — | — | — |
| GPQA Diamond | 82.0 | 86.8 | B +4.8 pts |
| Humanity's Last Exam | 29.3 | 30.1 | B +0.8 pts |
| AIME 2025 | — | — | — |
| MATH-500 | — | — | — |
| Humanity's Last Exam (with tools) | — | — | — |
| AA-Omniscience Accuracy | 26.3 | 23.7 | A +2.6 pts |
| AA-LCR v1.1 | 75.7 | 73.7 | A +2.0 pts |
| CritPt | 2.0 | 4.6 | B +2.6 pts |
| MMMU-Pro | — | — | — |
| IFBench | 72.3 | 76.3 | B +4.0 pts |
| Chartography | — | — | — |
| Chartography (With Tools) | — | — | — |
| Coding | |||
| SWE-bench Verified | 72.8 | — | — |
| SWE-bench Pro | — | — | — |
| SWE-bench Multilingual | 69.7 | — | — |
| LiveCodeBench | — | — | — |
| Terminal-Bench 2.1 | — | 61.8 | — |
| Aider Polyglot | — | — | — |
| Terminal-Bench 3 | — | — | — |
| BigCodeBench | — | — | — |
| SciCode | 46.2 | 44.8 | A +1.4 pts |
| CursorBench | — | — | — |
| SWE-Rebench | 53.3 | — | — |
| NL2Repo-Bench | — | — | — |
| DeepSWE | — | — | — |
| WebDev Arena | 1,436 | 1,508 | B +72 Elo |
| Terminal-Bench 4.0 | — | — | — |
| CursorBench 4.0 | — | — | — |
| FrontierCode v1.1 (Main) | — | — | — |
| Arena | |||
| LMArena Elo | 1,458 | 1,466 | B +8 Elo |
GLM-5
- Context
- 200K / 33K out
- Parameters
- —
- Price
- $0.40 / $1.20
- Speed
- 80 tok/s · 0.35s TTFT
- Modalities
- text → text
- License
- MIT
GLM-5 open release - solid all-rounder before the 5.2 bump.
GLM-5.1
- Context
- 203K / 128K out
- Parameters
- 744B (40B act.)
- Price
- $1.40 / $4.40
- Speed
- —
- Modalities
- text → text
- License
- MIT
Zhipu's April 2026 refinement of the GLM-5 744B MoE (40B active) flagship for agentic engineering — MIT open weights, ~200K context, 128K max output. Benchmark figures here are Artificial Analysis and LMArena leaderboard snapshots, not Zhipu's own launch numbers.
Benchmark charts
Winner bars are emphasized. Per-benchmark deltas sit above each chart.