SWE-bench & software engineering
Compare SWE-bench Verified, SWE-Pro, SWE-Atlas, and Terminal-Bench scores for coding-focused model selection.
LLM Leaderboard
LLM leaderboard for SWE-bench, coding agents, benchmarks, and API pricing.
Compare GPT-5.5, Claude Opus, Gemini, Composer, DeepSeek, and frontier LLMs by SWE-bench, GPQA, MATH-500, HLE, Terminal-Bench, coding agent index, token pricing, and AskClash RWT scores. Compare GPT vs Claude vs Gemini, open-weight models, and the newest frontier releases in one benchmark table.
#1 Claude Mythos/Fable 5Current overall leader with score 96.7 across public benchmark cells and AskClash RWT.
SWE-bench & coding agentsTrack SWE-bench, SWE-Pro, SWE-Atlas, Terminal-Bench, and Artificial Analysis coding agent index scores side by side.
LLM pricing & contextSee input/output token pricing, context window, and billing access path for each model row.
Top LLM benchmark rankings
These ranked rows are cached from the live AskClash leaderboard. Open any model for SWE-bench breakdowns, benchmark cells, pricing, and comparison links.
| # | Model | Creator | Overall | SWE | Price in/out |
|---|---|---|---|---|---|
| 1 | Claude Mythos/Fable 5 | Anthropic | 96.7 | 95.5 | $10.0 / $50.0 |
| 2 | Claude Opus 4.8 (Adaptive) | Anthropic | 90.7 | 88.6 | $5.00 / $25.0 |
| 3 | GPT-5.5 xHigh | OpenAI | 87.3 | — | $5.00 / $30.0 |
| 4 | Claude Opus 4.7 (Adaptive) | Anthropic | 82.4 | 87.6 | $5.00 / $25.0 |
| 5 | GPT-5.4 xHigh | OpenAI | 73.8 | — | $2.50 / $15.0 |
| 6 | Gemini 3.5 Flash High | 69.7 | — | $1.50 / $9.00 | |
| 7 | Kimi K2.7 Code | Moonshot AI | 69.0 | 80.2 | $0.95 / $4.00 |
| 8 | Gemini 3.1 Pro | 66.9 | 80.6 | $2.00 / $12.0 | |
| 9 | Claude Opus 4.6 (Adaptive) | Anthropic | 66.8 | 80.8 | $5.00 / $25.0 |
| 10 | Kimi K2.6 Thinking | Moonshot AI | 63.2 | 80.2 | $0.95 / $4.00 |
| 11 | Qwen3.7 Max | Alibaba | 58.4 | 80.4 | $2.50 / $7.50 |
| 12 | Qwen3.7 Plus | Alibaba | 58.4 | 77.7 | $0.40 / $1.60 |
| 13 | Composer 2.5 | Cursor | 57.9 | — | $0.50 / $2.50 |
| 14 | MiniMax-M3 | MiniMax | 52.4 | 80.5 | $0.30 / $1.20 |
| 15 | DeepSeek V4 Pro (Max) | DeepSeek | 45.6 | 80.6 | $1.74 / $3.48 |
Benchmarks and search topics covered
This page targets common LLM comparison queries: best coding LLM, SWE-bench leaderboard, GPT-5.5 vs Claude, Gemini benchmark scores, LLM API pricing comparison, and frontier model rankings.
Reasoning & knowledge
Track GPQA, MATH-500, HLE, ARC-AGI-2, Tau2, and multimodal benchmarks like MMMU-Pro when providers publish them.
AskClash RWT
Real World Testing adds a hands-on quality signal on top of public benchmark tables so rankings reflect practical use, not only vendor cards.
Frequently asked questions
Which models are on the leaderboard?
Frontier proprietary models, adaptive variants, and major open-weight releases from OpenAI, Anthropic, Google, Meta, DeepSeek, Alibaba, Moonshot, Z.ai, and Cursor when benchmark data is available.
Is this the same as Artificial Analysis or LMSYS?
AskClash combines multiple public benchmark sources into one weighted table with pricing, context, access path, and AskClash RWT instead of showing a single vendor index alone.
How often do rankings update?
Benchmark snapshots refresh from cached collector data. Reload the live leaderboard for the newest model rows and scores.
The interactive leaderboard loads in the browser. This crawlable page gives search engines stable copy for LLM leaderboard, benchmark comparison, and model ranking intent.