Compare · ModelsLive · 3 picked · head to head
DeepSeek V3.2 Speciale vs GLM 5 vs Step 3.5 Flash
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
DeepSeek V3.2 Speciale wins on 5/10 benchmarks
DeepSeek V3.2 Speciale wins 5 of 10 shared benchmarks. Leads in math · knowledge.
Category leads
math·DeepSeek V3.2 Specialeknowledge·DeepSeek V3.2 Specialelanguage·GLM 5coding·GLM 5speed·Step 3.5 Flasharena·GLM 5
Hype vs Reality
Attention vs performance
DeepSeek V3.2 Speciale
#6 by perf·#5 by attention
GLM 5
#55 by perf·#27 by attention
Step 3.5 Flash
#9 by perf·#11 by attention
Best value
Step 3.5 Flash
3.9x better value than DeepSeek V3.2 Speciale
DeepSeek V3.2 Speciale
97.8 pts/$
$0.80/M
GLM 5
45.7 pts/$
$1.26/M
Step 3.5 Flash
384.5 pts/$
$0.20/M
Vendor risk
Mixed exposure
One or more vendors flagged
DeepSeek
$3.4B·Tier 1
z-ai
private · undisclosed
StepFun
$5.0B·Tier 1
Head to head
10 benchmarks · 3 models
DeepSeek V3.2 SpecialeGLM 5Step 3.5 Flash
OpenCompass · AIME2025
DeepSeek V3.2 Speciale leads by +0.2
DeepSeek V3.2 Speciale
96.0
GLM 5
95.8
Step 3.5 Flash
95.7
OpenCompass · GPQA-Diamond
DeepSeek V3.2 Speciale leads by +1.4
DeepSeek V3.2 Speciale
86.7
GLM 5
85.3
Step 3.5 Flash
83.7
OpenCompass · HLE
DeepSeek V3.2 Speciale leads by +0.5
DeepSeek V3.2 Speciale
28.6
GLM 5
28.1
Step 3.5 Flash
21.6
OpenCompass · IFEval
DeepSeek V3.2 Speciale
91.7
GLM 5
93.2
Step 3.5 Flash
93.2
OpenCompass · LiveCodeBenchV6
GLM 5 leads by +2.3
DeepSeek V3.2 Speciale
80.9
GLM 5
86.2
Step 3.5 Flash
83.9
OpenCompass · MMLU-Pro
DeepSeek V3.2 Speciale leads by +0.3
DeepSeek V3.2 Speciale
85.5
GLM 5
85.2
Step 3.5 Flash
83.5
Artificial Analysis · Agentic Index
Step 3.5 Flash leads by +52.0
Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?"
DeepSeek V3.2 Speciale
0.0
Step 3.5 Flash
52.0
Artificial Analysis · Coding Index
DeepSeek V3.2 Speciale leads by +6.3
Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads.
DeepSeek V3.2 Speciale
37.9
Step 3.5 Flash
31.6
Artificial Analysis · Quality Index
Step 3.5 Flash leads by +8.4
DeepSeek V3.2 Speciale
29.4
Step 3.5 Flash
37.8
Chatbot Arena Elo · Overall
GLM 5 leads by +64.2
GLM 5
1455.6
Step 3.5 Flash
1391.4
Full benchmark table
| Benchmark | DeepSeek V3.2 Speciale | GLM 5 | Step 3.5 Flash |
|---|---|---|---|
OpenCompass · AIME2025 | 96.0 | 95.8 | 95.7 |
OpenCompass · GPQA-Diamond | 86.7 | 85.3 | 83.7 |
OpenCompass · HLE | 28.6 | 28.1 | 21.6 |
OpenCompass · IFEval | 91.7 | 93.2 | 93.2 |
OpenCompass · LiveCodeBenchV6 | 80.9 | 86.2 | 83.9 |
OpenCompass · MMLU-Pro | 85.5 | 85.2 | 83.5 |
Artificial Analysis · Agentic Index Artificial Analysis Agentic Index · a composite score measuring how well a model performs in agentic workflows · multi-step tool use, planning, error recovery, and autonomous task completion. Aggregates results from multiple agentic benchmarks including SWE-bench, tool-use tests, and planning evaluations. The canonical single-number metric for "how good is this model as an agent?" | 0.0 | — | 52.0 |
Artificial Analysis · Coding Index Artificial Analysis Coding Index · a composite score that aggregates performance across multiple coding benchmarks into a single index. Tracks code generation quality, debugging ability, multi-language competence, and real-world software engineering tasks. Used by Artificial Analysis to rank model coding capability in a normalized, comparable format. Useful for developers choosing between models for coding-heavy workloads. | 37.9 | — | 31.6 |
Artificial Analysis · Quality Index | 29.4 | — | 37.8 |
Chatbot Arena Elo · Overall | — | 1455.6 | 1391.4 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.40 | $1.20 | 164K tokens (~82 books) | $6.00 | |
| $0.60 | $1.92 | 203K tokens (~101 books) | $9.30 | |
| $0.10 | $0.30 | 262K tokens (~131 books) | $1.50 |