Dataset Preview
Duplicate
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed because of a cast error
Error code:   DatasetGenerationCastError
Exception:    DatasetGenerationCastError
Message:      An error occurred while generating the dataset

All the data files must have the same columns, but at some point there are 15 new columns ({'tool_use', 'batch_api', 'fine_tuning', 'file_upload', 'safety_filters', 'json_mode', 'multilingual', 'system_prompt', 'streaming', 'function_calling', 'web_search', 'vision', 'image_generation', 'embedding', 'code_execution'}) and 8 missing columns ({'multilingual_rank', 'reasoning_rank', 'humaneval_score', 'mmlu_score', 'coding_rank', 'arena_elo', 'math_score', 'overall_tier'}).

This happened while the csv dataset builder was generating data using

hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026/llm-features-2026.csv (at revision f777c54b5309a30a76fcc6069e2a0037f33f4b6f), [/tmp/hf-datasets-cache/medium/datasets/88909843474045-config-parquet-and-info-ComparEdge-llm-api-benchm-6872e516/hub/datasets--ComparEdge--llm-api-benchmark-matrix-2026/snapshots/f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-benchmarks-2026.csv (origin=hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026@f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-benchmarks-2026.csv), /tmp/hf-datasets-cache/medium/datasets/88909843474045-config-parquet-and-info-ComparEdge-llm-api-benchm-6872e516/hub/datasets--ComparEdge--llm-api-benchmark-matrix-2026/snapshots/f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-features-2026.csv (origin=hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026@f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-features-2026.csv), /tmp/hf-datasets-cache/medium/datasets/88909843474045-config-parquet-and-info-ComparEdge-llm-api-benchm-6872e516/hub/datasets--ComparEdge--llm-api-benchmark-matrix-2026/snapshots/f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-rate-limits-2026.csv (origin=hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026@f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-rate-limits-2026.csv)]

Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1800, in _prepare_split_single
                  writer.write_table(table)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 765, in write_table
                  self._write_table(pa_table, writer_batch_size=writer_batch_size)
                File "/usr/local/lib/python3.12/site-packages/datasets/arrow_writer.py", line 773, in _write_table
                  pa_table = table_cast(pa_table, self._schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2281, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2227, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              provider: string
              model: string
              vision: bool
              function_calling: bool
              json_mode: bool
              streaming: bool
              batch_api: bool
              fine_tuning: bool
              system_prompt: bool
              tool_use: bool
              image_generation: bool
              code_execution: bool
              web_search: bool
              file_upload: bool
              embedding: bool
              multilingual: bool
              safety_filters: bool
              -- schema metadata --
              pandas: '{"index_columns": [{"kind": "range", "name": null, "start": 0, "' + 2245
              to
              {'provider': Value('string'), 'model': Value('string'), 'mmlu_score': Value('float64'), 'humaneval_score': Value('float64'), 'math_score': Value('float64'), 'arena_elo': Value('int64'), 'coding_rank': Value('int64'), 'reasoning_rank': Value('int64'), 'multilingual_rank': Value('int64'), 'overall_tier': Value('string')}
              because column names don't match
              
              During handling of the above exception, another exception occurred:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1347, in compute_config_parquet_and_info_response
                  parquet_operations = convert_to_parquet(builder)
                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 980, in convert_to_parquet
                  builder.download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 882, in download_and_prepare
                  self._download_and_prepare(
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 943, in _download_and_prepare
                  self._prepare_split(split_generator, **prepare_split_kwargs)
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1646, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1802, in _prepare_split_single
                  raise DatasetGenerationCastError.from_cast_error(
              datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
              
              All the data files must have the same columns, but at some point there are 15 new columns ({'tool_use', 'batch_api', 'fine_tuning', 'file_upload', 'safety_filters', 'json_mode', 'multilingual', 'system_prompt', 'streaming', 'function_calling', 'web_search', 'vision', 'image_generation', 'embedding', 'code_execution'}) and 8 missing columns ({'multilingual_rank', 'reasoning_rank', 'humaneval_score', 'mmlu_score', 'coding_rank', 'arena_elo', 'math_score', 'overall_tier'}).
              
              This happened while the csv dataset builder was generating data using
              
              hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026/llm-features-2026.csv (at revision f777c54b5309a30a76fcc6069e2a0037f33f4b6f), [/tmp/hf-datasets-cache/medium/datasets/88909843474045-config-parquet-and-info-ComparEdge-llm-api-benchm-6872e516/hub/datasets--ComparEdge--llm-api-benchmark-matrix-2026/snapshots/f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-benchmarks-2026.csv (origin=hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026@f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-benchmarks-2026.csv), /tmp/hf-datasets-cache/medium/datasets/88909843474045-config-parquet-and-info-ComparEdge-llm-api-benchm-6872e516/hub/datasets--ComparEdge--llm-api-benchmark-matrix-2026/snapshots/f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-features-2026.csv (origin=hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026@f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-features-2026.csv), /tmp/hf-datasets-cache/medium/datasets/88909843474045-config-parquet-and-info-ComparEdge-llm-api-benchm-6872e516/hub/datasets--ComparEdge--llm-api-benchmark-matrix-2026/snapshots/f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-rate-limits-2026.csv (origin=hf://datasets/ComparEdge/llm-api-benchmark-matrix-2026@f777c54b5309a30a76fcc6069e2a0037f33f4b6f/llm-rate-limits-2026.csv)]
              
              Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

provider
string
model
string
mmlu_score
float64
humaneval_score
float64
math_score
float64
arena_elo
int64
coding_rank
int64
reasoning_rank
int64
multilingual_rank
int64
overall_tier
string
OpenAI
gpt-4o
88.7
90.2
76.6
1,287
2
3
2
S
OpenAI
gpt-4o-mini
82
87
70.2
1,140
5
8
5
A
OpenAI
o1
92.3
92.4
96.4
1,350
1
1
3
S+
OpenAI
o3-mini
86.9
89.5
94.8
1,310
3
2
6
S
Anthropic
claude-sonnet-4
88.5
93.7
78.3
1,295
1
3
2
S
Anthropic
claude-haiku-3.5
79.8
88.1
69.5
1,160
6
7
4
A
Anthropic
claude-opus-4
90.2
95.1
82.7
1,330
1
2
1
S+
Google
gemini-2.5-pro
90.8
89.8
86.2
1,340
2
1
1
S+
Google
gemini-2.5-flash
85.1
86.4
78
1,250
4
4
3
S
Google
gemini-2.0-flash
82.5
84.2
72.1
1,190
7
6
4
A
Meta
llama-3.3-70b
82
81.7
68
1,180
8
8
7
A
Meta
llama-3.1-405b
86.1
84.3
73.8
1,230
5
5
5
A+
Meta
llama-4-maverick
87.5
86
77.2
1,260
4
4
4
S
Mistral
mistral-large
84
82.5
72
1,200
6
6
3
A+
Mistral
mistral-small
78.5
79
65.3
1,120
9
9
6
B+
Mistral
codestral
75.2
90.8
58.4
1,170
2
10
8
A
DeepSeek
deepseek-v3
87.1
82.6
84
1,280
4
3
5
S
DeepSeek
deepseek-r1
89.5
85.2
94.3
1,320
3
1
6
S+
Cohere
command-r-plus
80.2
72.5
62.1
1,130
10
9
2
B+
Cohere
command-r
75.8
68.3
55.7
1,050
12
11
4
B
xAI
grok-2
85.5
83
71.5
1,220
5
5
6
A+
xAI
grok-3-mini
81
80.5
78.9
1,200
7
4
7
A
OpenAI
gpt-4o
null
null
null
null
null
null
null
null
OpenAI
gpt-4o-mini
null
null
null
null
null
null
null
null
OpenAI
o1
null
null
null
null
null
null
null
null
OpenAI
o3-mini
null
null
null
null
null
null
null
null
Anthropic
claude-sonnet-4
null
null
null
null
null
null
null
null
Anthropic
claude-haiku-3.5
null
null
null
null
null
null
null
null
Anthropic
claude-opus-4
null
null
null
null
null
null
null
null
Google
gemini-2.5-pro
null
null
null
null
null
null
null
null
Google
gemini-2.5-flash
null
null
null
null
null
null
null
null
Google
gemini-2.0-flash
null
null
null
null
null
null
null
null
Meta
llama-3.3-70b
null
null
null
null
null
null
null
null
Meta
llama-3.1-405b
null
null
null
null
null
null
null
null
Meta
llama-4-maverick
null
null
null
null
null
null
null
null
Mistral
mistral-large
null
null
null
null
null
null
null
null
Mistral
mistral-small
null
null
null
null
null
null
null
null
Mistral
codestral
null
null
null
null
null
null
null
null
DeepSeek
deepseek-v3
null
null
null
null
null
null
null
null
DeepSeek
deepseek-r1
null
null
null
null
null
null
null
null
Cohere
command-r-plus
null
null
null
null
null
null
null
null
Cohere
command-r
null
null
null
null
null
null
null
null
xAI
grok-2
null
null
null
null
null
null
null
null
xAI
grok-3-mini
null
null
null
null
null
null
null
null
OpenAI
gpt-4o
null
null
null
null
null
null
null
null
OpenAI
gpt-4o-mini
null
null
null
null
null
null
null
null
OpenAI
o1
null
null
null
null
null
null
null
null
OpenAI
o3-mini
null
null
null
null
null
null
null
null
Anthropic
claude-sonnet-4
null
null
null
null
null
null
null
null
Anthropic
claude-haiku-3.5
null
null
null
null
null
null
null
null
Anthropic
claude-opus-4
null
null
null
null
null
null
null
null
Google
gemini-2.5-pro
null
null
null
null
null
null
null
null
Google
gemini-2.5-flash
null
null
null
null
null
null
null
null
Google
gemini-2.0-flash
null
null
null
null
null
null
null
null
Meta
llama-3.3-70b
null
null
null
null
null
null
null
null
Meta
llama-3.1-405b
null
null
null
null
null
null
null
null
Meta
llama-4-maverick
null
null
null
null
null
null
null
null
Mistral
mistral-large
null
null
null
null
null
null
null
null
Mistral
mistral-small
null
null
null
null
null
null
null
null
Mistral
codestral
null
null
null
null
null
null
null
null
DeepSeek
deepseek-v3
null
null
null
null
null
null
null
null
DeepSeek
deepseek-r1
null
null
null
null
null
null
null
null
Cohere
command-r-plus
null
null
null
null
null
null
null
null
Cohere
command-r
null
null
null
null
null
null
null
null
xAI
grok-2
null
null
null
null
null
null
null
null
xAI
grok-3-mini
null
null
null
null
null
null
null
null

LLM Benchmark & Feature Matrix 2026

Which LLM is best at what? This dataset maps capabilities, performance, and limits of 22 major models.

Unlike pricing datasets, this focuses on what models can do — not just what they cost.

Files

File Description
llm-benchmarks-2026.csv MMLU, HumanEval, MATH, Arena ELO, coding/reasoning/multilingual rankings, tier (S+ to B)
llm-features-2026.csv 15 binary capabilities: vision, function calling, JSON mode, fine-tuning, tool use, web search, embeddings...
llm-rate-limits-2026.csv Free tier availability, RPM/TPM limits, batch discounts, cached input discounts

Models Covered

22 models from OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, xAI, Cohere

Use Cases

  • Model selection — Find models that support your required features (e.g., vision + function calling + fine-tuning)
  • Performance comparison — Which model scores highest on coding vs reasoning vs multilingual?
  • Rate limit planning — Can you stay within free tier? What are the paid RPM limits?
  • Tier analysis — S+ tier models vs A tier — is the premium worth it?

Key Insights

  • Only Google Gemini supports all 15 features (vision, search, embeddings, fine-tuning, code execution)
  • DeepSeek offers 90% cached input discount — massive savings for repetitive workloads
  • Groq has highest free tier RPM (30) with lowest latency
  • S+ tier models (o1, Claude Opus 4, Gemini 2.5 Pro, DeepSeek R1) all score >89 MMLU

Related

License

CC BY 4.0 — ComparEdge

Downloads last month
25