Aggregated AI Model Rankings
Overall AI Model Ranking
Compare current model versions across independent benchmarks, community preference data, and official pricing. Scores show relative capability, not model IQ or an absolute performance gap.
167 ranked12 core benchmarks
- Sources
- 2
- Benchmarks
- 668
- Models
- 393
- Updated
- 2026-08-11
| 01 | AN | 97.4 | Leading | Reasoning, Knowledge | $5 | $25 | Moderate 60% | ||
| 02 | AN | 95.8 | Leading | Knowledge, Coding | $10 | $50 | Limited 40% | ||
| 03 | OP | 95.1 | Leading | Math, Reasoning | $5 | $30 | High 80% | ||
| 04 | AN | 93.6 | Leading | Math, Reasoning | N/A | N/A | High 80% | N/A | |
| 05 | MA | 93.0 | Leading | Reasoning, Knowledge | $3 | $15 | Moderate 60% | ||
| 06 | AC | 89.3 | Strong | Instruction Following, Reasoning | $2 | $6 | High 80% | ||
| 07 | ME | 88.3 | Strong | Knowledge, Coding | $1.25 | $4.25 | Limited 40% | ||
| 08 | AN | 87.9 | Strong | Reasoning, Knowledge | $5 | $25 | Moderate 60% | ||
| 09 | OP | 87.6 | Strong | Reasoning, Math | $2 | $12 | High 80% | ||
| 10 | AN | 84.2 | Strong | Reasoning, Knowledge | $5 | $25 | Moderate 60% | ||
| 11 | AN | 84.0 | Strong | Knowledge, Coding | $2 | $10 | Moderate 60% | ||
| 12 | OP | 84.0 | Strong | Reasoning, Knowledge | $5 | $30 | High 80% | ||
| 13 | ME | 82.3 | Provisional | Coding | $1.25 | $4.25 | Limited 20% | ||
| 14 | XA | 82.2 | Provisional | Reasoning, Coding | $2 | $6 | Limited 40% | ||
| 15 | AN | 79.6 | Strong | Math, Knowledge | $5 | $25 | High 80% | ||
| 16 | GO | 79.3 | Provisional | Coding | $1.5 | $7.5 | Limited 20% | ||
| 17 | ZA | 79.3 | Strong | Reasoning, Math | $1.4 | $4.4 | High 80% | ||
| 18 | GO | 79.2 | Strong | Reasoning, Knowledge | $1.5 | $9 | Moderate 60% | ||
| 19 | AC | 78.8 | Strong | Math, Instruction Following | $2.5 | $7.5 | High 100% | ||
| 20 | BY | 78.7 | Strong | Reasoning, Knowledge | N/A | N/A | High 80% | ||
| 21 | DE | 78.3 | Provisional | Coding | $0.14 | $0.28 | Limited 20% | ||
| 22 | GO | 77.5 | Strong | Reasoning, Knowledge | $2 | $12 | Moderate 60% | ||
| 23 | OP | 77.1 | Strong | Reasoning, Math | $2.5 | $15 | High 80% | ||
| 24 | OP | 76.4 | Strong | Reasoning, Math | $0.20 | $1.2 | High 80% | ||
| 25 | OP | 75.8 | Provisional | Knowledge, Math | $30 | $180 | Limited 40% | ||
| 26 | MA | 75.5 | Strong | Reasoning, Math | $0.95 | $4 | High 80% | ||
| 27 | DE | 73.8 | Competitive | Reasoning, Knowledge | $0.435 | $0.87 | High 80% | ||
| 28 | AC | 73.8 | Competitive | Instruction Following, Reasoning | $0.50 | $3 | High 100% | ||
| 29 | ME | 73.2 | Competitive | Knowledge, Reasoning | N/A | N/A | Moderate 60% | ||
| 30 | BY | 73.1 | Competitive | Reasoning, Knowledge | N/A | N/A | High 80% |
Rank by capability
Each leaderboard sorts by its own capability measure, not by the overall score.
CodingClaude Sonnet 4.5Code generation and software engineeringReasoningClaude Opus 5Complex problems and logical reasoningMathGPT-5.6-SolMathematics and quantitative reasoningKnowledgeClaude MythosKnowledge coverage and factual answersInstruction FollowingQwen3.8 MaxReliable completion of requested tasksOpen ModelsOpen-weight models onlyCompare capability, licensing, and cost separately
Core benchmarks
A focused set of broad-coverage, current, score-eligible evaluations.
GPQAphysicsMMLU-ProlanguageAIME 2025mathSWE-Bench VerifiedreasoningMMLUlanguageHumanity's Last ExammathLiveCodeBenchreasoningMATHmathHumanEvalreasoningMMMU-PromultimodalIFEvalinstruction followingMMMUmultimodal
Browse all 668 benchmarksCapability, price, and runtime
These selection signals are shown together but remain separate. Price and runtime never change the capability score.
Capability vs. official API price
Only models with positive official pay-as-you-go token prices are shown.
LLMBoard ScoreOfficial API price blend (8:1 input/output), USD / 1M
Efficient frontier128 models with official PAYG prices
Capability vs. output speed
Each scored version uses its fastest tracked model-provider record. Output Speed is generated tokens received per second.
LLMBoard scoreOutput speed
Each mark is one scored model version92 records
Runtime data is provider-specific and remains independent from capability, evidence coverage, and price. Open the full runtime page for catalog latency, TTFT, observed throughput, and reliability.