LLM Comparison

SWE-2 vs GPT-5.6 Luna: benchmark scores, pricing & comparison.

Side-by-side SWE-2 vs GPT-5.6 Luna comparison across SWE-bench, GPQA, HLE, Terminal-Bench, coding agent scores, token pricing, context window, and AskClash RWT. Green marks the winner on each benchmark.

Verdict: SWE-2 ranks higher (#22 vs #37) and wins 6 of 10 shared benchmarks; SWE-2 costs less ($0/$0 vs $1.00/$6.00 per 1M tokens).

Open live leaderboardSWE-2 model pageGPT-5.6 Luna model page
Rank #22 vs #37AskClash overall scores 55.3 vs 45.4.
Pricing $0/$0 vs $1.00/$6.00Input and output token prices per 1M tokens when published.
Proprietary vs ProprietaryCognition vs OpenAI.

SWE-2 vs GPT-5.6 Luna benchmark comparison

Green cells highlight the winning model for each metric. Scores are cached from the AskClash LLM leaderboard snapshot.

MetricSWE-2GPT-5.6 Luna
Overall Score55.345.4
Leaderboard Rank#22#37
ACB60.056.1
RWT8.59.5
Coding Agent Index51.975.0
HLE43.5—
GPQA93.592.3
SWE-Pro—62.7
SWE-Atlas66.181.0
Terminal-Bench27.311.6
DeepSWE73.067.2
GDPval-AA1668.01592.0
MCP Atlas84.2—
Finance Agent54.455.0
CharXiv84.8—
MMMU-Pro81.678.4
ARC-AGI 2—59.5
MRCR—41.3
Input Price (per 1M tokens)$0$1.00
Output Price (per 1M tokens)$0$6.00
Context Window256K1M
Benchmarks Published1312

SWE-2 vs GPT-5.6 Luna head-to-head charts

SWE-2 leads 7 and GPT-5.6 Luna leads 4 of 11 shared benchmarks. SWE-2 is cheaper on a defined 1M-input / 200K-output workload. Charts show only benchmarks both models publish.

SWE-2GPT-5.6 Luna
Overall
55.3SWE-2
45.4GPT-5.6 Luna
ACB
60.0SWE-2
56.1GPT-5.6 Luna
RWT
8.5SWE-2
9.5GPT-5.6 Luna
Coding Agent Index
51.9SWE-2
75.0GPT-5.6 Luna
GPQA
93.5SWE-2
92.3GPT-5.6 Luna
SWE-Atlas
66.1SWE-2
81.0GPT-5.6 Luna
Terminal-Bench
27.3SWE-2
11.6GPT-5.6 Luna
DeepSWE
73.0SWE-2
67.2GPT-5.6 Luna
GDPval-AA
1668.0SWE-2
1592.0GPT-5.6 Luna
Finance Agent
54.4SWE-2
55.0GPT-5.6 Luna
MMMU-Pro
81.6SWE-2
78.4GPT-5.6 Luna
SWE-2
Input$0
Output$0
Workload$0.00
Context256K
GPT-5.6 Luna
Input$1.00
Output$6.00
Workload$2.20
Context1M

Workload = published cost of 1M input + 200K output tokens. Open the live leaderboard for interactive compare charts.

More SWE-2 and GPT-5.6 Luna comparisons

Explore how SWE-2 and GPT-5.6 Luna stack up against other top-ranked LLMs.

Claude Opus 5.5 vs SWE-2Claude Opus 5.5 vs GPT-5.6 LunaClaude Sonnet 5.5 vs SWE-2Claude Sonnet 5.5 vs GPT-5.6 LunaClaude Fable 5.1 vs SWE-2Claude Fable 5.1 vs GPT-5.6 LunaClaude Opus 5 vs SWE-2Claude Opus 5 vs GPT-5.6 LunaClaude Fable 5 vs SWE-2Claude Fable 5 vs GPT-5.6 Luna

How to read this comparison

Benchmark scores

Higher is better for all benchmark scores (SWE-bench, GPQA, HLE, Terminal-Bench, etc.). Green marks the model with the higher score.

Token pricing

Lower is better for input and output prices. Green marks the cheaper model per 1M tokens.

Coverage matters

Models with fewer disclosed benchmark cells may have inflated percentile scores. Check the benchmark cell count for context.

This comparison page is generated from the AskClash LLM leaderboard cache. Open the live leaderboard for real-time scores and interactive filtering.