MODEL DIRECTORY
Follow the model.
See the evidence.
Benchmark results, evaluation settings and disclosed training methods—collected in one place for each model.
3,335 reported models and checkpoint variants. Scores keep their original units and sources.
3,335 models · page 22 of 70
Anthropic
Claude Fable 5.1 (Claude Code) xHigh*
2 benchmarks · 2 observations
Explore results →Anthropic
Claude Haiku 4.5 (Thinking) (Terminus 2)
2 benchmarks · 2 observations
Explore results →Anthropic
Claude Opus 4.7 (Terminus 2)
2 benchmarks · 2 observations
Explore results →Anthropic
Claude Opus 4.8 (Terminus 2)
2 benchmarks · 2 observations
Explore results →Anthropic
Claude Opus 5.5 (CVP)
2 benchmarks · 2 observations
Explore results →Anthropic
Claude Sonnet 4.6 (Terminus 2)
2 benchmarks · 2 observations
Explore results →Anthropic
claude-2 (Mar 2024)
2 benchmarks · 2 observations
Explore results →Anthropic
claude-3-5-haiku-20241022
2 benchmarks · 3 observations
Explore results →Anthropic
claude-3-7-sonnet-thinking
2 benchmarks · 5 observations
Explore results →Anthropic
claude-3-sonnet-20240229
2 benchmarks · 2 observations
Explore results →Anthropic
claude-4-5-Sonnet
2 benchmarks · 2 observations
Explore results →Anthropic
claude-instant-1 (Mar 2024)
2 benchmarks · 2 observations
Explore results →Anthropic
Claude2
2 benchmarks · 5 observations
Explore results →Reported model
Code-13B
2 benchmarks · 2 observations
Explore results →Reported model
Code-290k-6.7B-Instruct
2 benchmarks · 2 observations
Explore results →Reported model
Code-33B
2 benchmarks · 2 observations
Explore results →Reported model
CodeGen2-16B
2 benchmarks · 2 observations
Explore results →Reported model
CodeGen2-1B
2 benchmarks · 2 observations
Explore results →Reported model
CodeGen2-3B
2 benchmarks · 2 observations
Explore results →Reported model
CodeGen2-7B
2 benchmarks · 2 observations
Explore results →Meta
codellama-13b / instruct
2 benchmarks · 3 observations
Explore results →Meta
codellama-34b / instruct
2 benchmarks · 3 observations
Explore results →Meta
codellama-7b / instruct
2 benchmarks · 3 observations
Explore results →DeepSeek
DeepSeek V2 Chat
2 benchmarks · 2 observations
Explore results →DeepSeek
DeepSeek V4 (Terminus 2)
2 benchmarks · 2 observations
Explore results →DeepSeek
DeepSeek-Coder-33B-base
2 benchmarks · 2 observations
Explore results →DeepSeek
DeepSeek-Math-7B-Instruct
2 benchmarks · 6 observations
Explore results →DeepSeek
DeepSeek-Math-7B-RL
2 benchmarks · 4 observations
Explore results →DeepSeek
Deepseek-R1-671B
2 benchmarks · 5 observations
Explore results →Mistral
Devstral 2
2 benchmarks · 2 observations
Explore results →Reported model
Falcon-instruct (40B)
2 benchmarks · 2 observations
Explore results →Reported model
FinQANet (BERT-base)
2 benchmarks · 4 observations
Explore results →Reported model
FinQANet (BERT-large)
2 benchmarks · 4 observations
Explore results →Reported model
FinQANet (RoBERTa-base)
2 benchmarks · 4 observations
Explore results →Reported model
FinQANet (RoBERTa-large)
2 benchmarks · 4 observations
Explore results →Reported model
FinQANet-Gold (RoBERTa-large)
2 benchmarks · 4 observations
Explore results →Reported model
FLAN-PaLM (540B)
2 benchmarks · 5 observations
Explore results →Reported model
GAMA-IT
2 benchmarks · 412 observations
Explore results →Gemini 1.5 Pro (0409 preview)
2 benchmarks · 6 observations
Explore results →Gemini 2.0 Flash Experimental
2 benchmarks · 2 observations
Explore results →Gemini 2.0 Pro Exp
2 benchmarks · 2 observations
Explore results →Gemini 2.0 Pro exp-02-05
2 benchmarks · 2 observations
Explore results →Gemini 2.5 Flash Preview (May 2025)
2 benchmarks · 2 observations
Explore results →Gemini 2.5 Pro Experimental (March 2025)
2 benchmarks · 2 observations
Explore results →Gemini 2.5 Pro Preview (May 06 2025)
2 benchmarks · 2 observations
Explore results →Gemini 2.5 Pro Preview 03-25
2 benchmarks · 2 observations
Explore results →Gemini 3 Flash (12/25) (Terminus 2)
2 benchmarks · 2 observations
Explore results →Gemini 3.1 Flash Lite Preview (Terminus 2)
2 benchmarks · 2 observations
Explore results →