Compare · ModelsLive · 2 picked · head to head
Grok 4.3 vs Qwen3.6 27B
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Grok 4.3 wins on 13/17 benchmarks
Grok 4.3 wins 13 of 17 shared benchmarks. Leads in knowledge · general · math.
Category leads
speed·Qwen3.6 27Bknowledge·Grok 4.3general·Grok 4.3math·Grok 4.3coding·Grok 4.3reasoning·Qwen3.6 27Blanguage·Grok 4.3
Hype vs Reality
Attention vs performance
Grok 4.3
#68 by perf·#15 by attention
Qwen3.6 27B
#65 by perf·#2 by attention
Best value
Qwen3.6 27B
1.3x better value than Grok 4.3
Grok 4.3
30.1 pts/$
$1.88/M
Qwen3.6 27B
37.7 pts/$
$1.51/M
Vendor risk
Who is behind the model
xAI
$250.0B·Tier 1
Alibaba (Qwen)
$293.0B·Tier 1
Head to head
17 benchmarks · 2 models
Grok 4.3Qwen3.6 27B
Artificial Analysis · Agentic Index
Qwen3.6 27B leads by +2.9
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?"
Grok 4.3
24.1
Qwen3.6 27B
27.0
Artificial Analysis · Coding Index
Qwen3.6 27B leads by +11.5
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.
Grok 4.3
42.3
Qwen3.6 27B
53.7
Artificial Analysis · Quality Index
Grok 4.3 leads by +0.5
Grok 4.3
37.6
Qwen3.6 27B
37.0
Chess Puzzles
Grok 4.3 leads by +3.2
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Grok 4.3
21.1
Qwen3.6 27B
17.9
Dtbench
Grok 4.3 leads by +20.9
Grok 4.3
84.5
Qwen3.6 27B
63.5
FrontierMath-Tiers-1-3-v2-Private
Grok 4.3 leads by +7.7
Grok 4.3
42.8
Qwen3.6 27B
35.1
GPQA diamond
Grok 4.3 leads by +4.0
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Grok 4.3
85.1
Qwen3.6 27B
81.1
LiveBench · Agentic Coding
Grok 4.3
50.0
Qwen3.6 27B
50.0
LiveBench · Coding
Qwen3.6 27B leads by +1.9
Grok 4.3
69.9
Qwen3.6 27B
71.8
LiveBench · Data Analysis
Qwen3.6 27B leads by +14.7
Grok 4.3
55.8
Qwen3.6 27B
70.4
LiveBench · If
Grok 4.3 leads by +9.5
Grok 4.3
62.8
Qwen3.6 27B
53.2
LiveBench · Language
Grok 4.3 leads by +10.3
Grok 4.3
73.6
Qwen3.6 27B
63.3
LiveBench · Mathematics
Grok 4.3 leads by +4.5
Grok 4.3
84.3
Qwen3.6 27B
79.9
LiveBench · Overall
Grok 4.3 leads by +1.2
Grok 4.3
66.7
Qwen3.6 27B
65.6
LiveBench · Reasoning
Grok 4.3 leads by +0.5
Grok 4.3
70.8
Qwen3.6 27B
70.3
Lmca
Grok 4.3 leads by +4.5
Grok 4.3
45.1
Qwen3.6 27B
40.6
OTIS Mock AIME 2024-2025
Grok 4.3 leads by +2.2
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Grok 4.3
93.3
Qwen3.6 27B
91.1
Full benchmark table
| Benchmark | Grok 4.3 | Qwen3.6 27B |
|---|---|---|
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?" | 24.1 | 27.0 |
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. | 42.3 | 53.7 |
Artificial Analysis · Quality Index | 37.6 | 37.0 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 21.1 | 17.9 |
Dtbench | 84.5 | 63.5 |
FrontierMath-Tiers-1-3-v2-Private | 42.8 | 35.1 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 85.1 | 81.1 |
LiveBench · Agentic Coding | 50.0 | 50.0 |
LiveBench · Coding | 69.9 | 71.8 |
LiveBench · Data Analysis | 55.8 | 70.4 |
LiveBench · If | 62.8 | 53.2 |
LiveBench · Language | 73.6 | 63.3 |
LiveBench · Mathematics | 84.3 | 79.9 |
LiveBench · Overall | 66.7 | 65.6 |
LiveBench · Reasoning | 70.8 | 70.3 |
Lmca | 45.1 | 40.6 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 93.3 | 91.1 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $1.25 | $2.50 | 1.0M tokens (~500 books) | $15.63 | |
| $0.32 | $2.70 | 262K tokens (~131 books) | $9.15 |
People also compared