GLM-5V-Turbo vs Spark-X2.6-4B

Change models
Spark-X2.6-4BSparkLLM
IndexAll categories, out of 100.
47.9 ±8.6
48.2 ±5.2 best
GapIndex points below the top.
−0.3 · no clear gap
Top
CoverageIndex weight that has results.
65%
85% best
SpeedOutput tokens per second.
27/s
—
EloLMArena rating and rank.
—
—
ReleasedMonth of release.
2026-04
2026-10
Input / 1MUS dollars per 1M input tokens.
$1.20
—
Output / 1MUS dollars per 1M output tokens.
$4.00
—
1K chats2K tokens in, 500 out, each.
$4.40
—
1K documents20K tokens in, 1K out, each.
$28.00
—
1K agent tasks200K tokens in, 20K out, each.
$320.00
—
ContextMaximum tokens in one request.
203K
1M best
AgenticMulti-step tasks with tools.
50.4 best
48.5
CodingCode writing and repair.
—
48.4
ReasoningLogic problems and puzzles.
46.3
48.2 best
MultimodalTasks with images and text.
49.5
—
KnowledgeFacts and expert knowledge.
46.9 best
43.1
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
45.8
55.0 best
MathMath problems.
—
49.9

Benchmarks

3 shared
Only GLM-5V-Turbo: 7Only Spark-X2.6-4B: 17
GLM-5V-Turbo
Spark-X2.6-4B
70.3% best107 / 211
67.3%124 / 211
98.5% best3 / 143
86.9%47 / 143
Claw-EvalAgentic
53.8%36 / 42
69.9% best10 / 42