| 1 | Claude Fable 5.2 | 11 | 16 | 19 | SimpleBench, LiveBench, FrontierMath, SWE Atlas⦠|
|---|
| 2 | Gemini 4 Carbon | 6 | 11 | 18 | CritPt, MathArena, Vending-Bench 2, APEX-Agents⦠|
|---|
| 3 | Claude Opus 5.5 | 4 | 9 | 21 | Terminal-Bench, GDPval-AA, Design Arena, Scale Fortress |
|---|
| 4 | GPT-6 Astra | 3 | 11 | 20 | Humanity's Last Exam (HLE), ARC-AGI (ARC Prize), EuroEval |
|---|
| 5 | Claude Opus 5 | 3 | 5 | 19 | EQ-Bench, SWE-Bench Pro, OSWorld 2.0 |
|---|
| 6 | DeepSeek V4.1 Flash | 2 | 3 | 13 | OpenRouter Rankings, Language models (our test) |
|---|
| 7 | Kimi K3 | 2 | 3 | 9 | MCPMark, OCR (our test) |
|---|
| 8 | DeepSeek-V4-Flash | 2 | 3 | 6 | Language models (our test), Language models (our test) |
|---|
| 9 | Qwen3.5-397B-A17B | 2 | 3 | 5 | OCR (our test), OCR (our test) |
|---|
| 10 | Qwen3-ASR-1.7B | 2 | 3 | 3 | ASR (our test), ASR (our test) |
|---|
| 11 | gemini-3.1-flash-lite | 2 | 2 | 4 | ASR (our test), OCR (our test) |
|---|
| 12 | GPT-6.1 Sol | 1 | 5 | 19 | MultiChallenge |
|---|
| 13 | Kimi K2.6 | 1 | 3 | 6 | OCR (our test) |
|---|
| 14 | DeepSeek-V4-Pro | 1 | 3 | 6 | Reasoning level (reasoning_effort) (our test) |
|---|
| 15 | Claude Fable 5 | 1 | 2 | 14 | SWE-rebench |
|---|
| 16 | GPT-5.6 Sol | 1 | 2 | 14 | Context Arena (MRCR) |
|---|
| 17 | GLM 5.3 Flash | 1 | 2 | 7 | Toolathlon (Tool Decathlon) |
|---|
| 18 | OpenAI gpt-live-transcribe | 1 | 2 | 2 | ASR (our test) |
|---|
| 19 | Qwen3-VL 235B-A22B | 1 | 1 | 5 | OCR (our test) |
|---|
| 20 | Anthropic Sonnet 5 | 1 | 1 | 4 | F5 Labs CASI & ARS Leaderboard |
|---|
| 21 | gpt-transcribe | 1 | 1 | 2 | ASR (our test) |
|---|
| 22 | Muse-Glimmer-30B | 1 | 1 | 2 | CPTU-Bench |
|---|
| 23 | Qwen3-Embedding-0.6B | 1 | 1 | 2 | Embeddings (our test) |
|---|
| 24 | Mistral-Small-3.2-24B-Instruct | 1 | 1 | 2 | OCR (our test) |
|---|
| 25 | Intelligence-Indeed Agent | 1 | 1 | 1 | OSWorld (OSWorld-Verified) |
|---|