GLM-5 vs Qwen3.5-122B-A10B

Change models
GLM-5Z.AI
IndexAll categories, out of 100.
51.0 ±3.6
51.2 ±4.3 best
GapIndex points below the top.
−0.2 · no clear gap
Top
CoverageIndex weight that has results.
90%
90%
SpeedOutput tokens per second.
43/s
129/s best
EloLMArena rating and rank.
1446 best
1418
ReleasedMonth of release.
2026-02
2026-03
Input / 1MUS dollars per 1M input tokens.
$1.00
$0.26 best
Output / 1MUS dollars per 1M output tokens.
$3.20
$2.08 best
1K chats2K tokens in, 500 out, each.
$3.60
$1.56 best
1K documents20K tokens in, 1K out, each.
$23.20
$7.28 best
1K agent tasks200K tokens in, 20K out, each.
$264.00
$93.60 best
ContextMaximum tokens in one request.
205K
262K best
AgenticMulti-step tasks with tools.
50.9 best
50.8
CodingCode writing and repair.
52.9 best
49.8
ReasoningLogic problems and puzzles.
47.9
49.7 best
MultimodalTasks with images and text.
—
53.1
KnowledgeFacts and expert knowledge.
51.5
52.4 best
MultilingualTasks in many languages.
52.0 best
51.2
InstructionTasks with strict prompt rules.
52.4
54.7 best
MathMath problems.
51.7
—

Benchmarks

15 shared
Only GLM-5: 48Only Qwen3.5-122B-A10B: 13
GLM-5
Qwen3.5-122B-A10B
29.3% best82 / 216
25.2%95 / 216
75.7%88 / 211
76.3% best84 / 211
2.0% best102 / 207
0.6%127 / 207
82.0%88 / 193
85.7% best69 / 193
86.0%50 / 144
86.6% best47 / 144
98.2% best7 / 143
93.6%28 / 143
72.3%36 / 140
75.7% best18 / 140
77.8% best31 / 114
72.0%62 / 114
85.7%15 / 73
86.7% best9 / 73
92.6%16 / 69
93.4% best14 / 69
BrowseCompAgentic
62.0%51 / 63
63.8% best50 / 63
56.2% best24 / 40
49.4%30 / 40
LongBench v2Reasoning
60.8% best13 / 25
60.2%16 / 25
66.8%13 / 21
67.1% best12 / 21
MMLU-ProXMultilingual
83.1% best6 / 16
82.2%9 / 16