ModelsCompareBest forBenchmarksStatusPricingAPI
Compare / head-to-head

DeepSeek V3.2vsGLM-5

GLM-5 leads 5 of 5 shared benchmarks. GLM-5 has the larger context window (200K tokens).

Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.

Shared benchmarks
0 – 5
GLM-5 leads
Cheaper per token
list price, input + output
Larger context
GLM-5
200K tokens
Provider uptime (30d)
100% · —
DeepSeek · Z.ai
DeepSeek V3.2
DeepSeek · released 2025-12-01
text
Context
131K
Max out
66K
Input /1M
Output /1M
Cached /1M
Scores
8 · 7 core
GLM-5
Z.ai
text
Context
200K
Max out
131K
Input /1M
$1
Output /1M
$3.2
Cached /1M
$0.2
Scores
8 · 7 core
Quality

Benchmark matrix

BenchmarkDeepSeek V3.2GLM-5ΔEdge
Reported by both · 5
SWE-bench Verified73.1% 77.8% -4.7 ptGLM-5
GPQA Diamond82.4% 87.8% epoch-5.4 ptGLM-5
AIME 202593.1% 96.7% matharena-3.6 ptGLM-5
AIME 202694.2% matharena96.7% matharena-2.5 ptGLM-5
LMArena Elo1424.8 1446.3 -21.5GLM-5
Only DeepSeek V3.2 reports · 2
MMLU-Pro85% not reported
Humanity's Last Exam (no tools)25.1% not reported
Only GLM-5 reports · 3
OTIS Mock AIME 2024-2025 (Epoch AI run)not reported80% epoch
SWE-bench Verified (Epoch AI run)not reported72.1% epoch
Terminal-Bench 2.0not reported56.2%
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is DeepSeek V3.2 minus GLM-5 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing

Side by side

SpecDeepSeek V3.2GLM-5Edge
Context window131K tokens200K tokensGLM-5
Max output66K tokens131K tokensGLM-5
Input price / 1M$1
Output price / 1M$3.2
Cached input / 1M$0.2
Input + output / 1M
Lower is cheaper. List prices; batch, tool and regional fees excluded.
$4.2
ModalitiestexttextTie
Released2025-12-01
Cited benchmark scores88
Reliability

Provider status

All providers
Built by Respan
Which one wins on your data?

Public benchmarks are a starting point. Run DeepSeek V3.2 and GLM-5 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.