Evals#
gptme provides LLMs with a wide variety of tools, but how well do models make use of them? Which tasks can they complete, and which ones do they struggle with? How far can they get on their own, without any human intervention?
To answer these questions, we have created an evaluation suite that tests the capabilities of LLMs on a wide variety of tasks.
Note
The evaluation suite is still tiny and under development, but the eval harness is fully functional.
Recommended Model#
The recommended model is Claude Sonnet 4.5 (anthropic/claude-sonnet-4-5 and openrouter/anthropic/claude-sonnet-4-5) for its:
Strong agentic capabilities
Strong coder capabilities
Strong performance across all tool types and formats
Reasoning capabilities
Vision & computer use capabilities
Decent alternatives include:
GPT-4o (
openai/gpt-4o)Llama 3.1 405B (
openrouter/meta-llama/llama-3.1-405b-instruct)DeepSeek V3 (
deepseek/deepseek-chat)DeepSeek R1 (
deepseek/deepseek-reasoner)
Usage#
You can run the simple hello eval like this:
gptme-eval hello --model anthropic/claude-sonnet-4-5
However, we recommend running it in Docker to improve isolation and reproducibility:
make build-docker
docker run \
-e "ANTHROPIC_API_KEY=<your api key>" \
-v $(pwd)/eval_results:/app/eval_results \
gptme-eval hello --model anthropic/claude-sonnet-4-5
Available Evals#
The current evaluations test basic tool use in gptme, such as the ability to: read, write, patch files; run code in ipython, commands in the shell; use git and create new projects with npm and cargo. It also has basic tests for web browsing and data extraction.
Results#
Here are the results of the evals we have run so far:
$ gptme-eval eval_results/*/eval_results.csv
Model hello hello-patch hello-ask prime100 init-git init-rust whois-superuserlabs-ceo
--------------------------------------------- ---------------- ---------------- ---------------- ---------------- ---------------- ------------- -------------------------
anthropic/claude-3-5-haiku-20241022 ✅ 55/57 473tk ✅ 56/57 386tk ✅ 56/57 468tk ✅ 56/57 912tk ✅ 54/56 922tk ❓ N/A ✅ 1/1 2619tk
anthropic/claude-3-5-haiku-20241022@markdown ✅ 11/12 373tk ✅ 12/12 332tk ✅ 12/12 407tk ✅ 11/12 758tk ✅ 11/12 830tk ❓ N/A ❓ N/A
anthropic/claude-3-5-haiku-20241022@tool 🔶 59/118 285tk 🔶 60/118 323tk 🔶 60/118 427tk ❌ 12/118 589tk 🔶 65/118 630tk ❓ N/A ❓ N/A
anthropic/claude-3-5-haiku-20241022@xml 🔶 59/106 296tk 🔶 60/106 322tk 🔶 43/106 440tk ❌ 11/106 705tk 🔶 54/106 621tk ❓ N/A ❓ N/A
anthropic/claude-3-5-sonnet-20240620 ✅ 34/34 630tk ✅ 34/34 542tk ✅ 34/34 710tk ✅ 34/34 924tk ✅ 34/34 1098tk ✅ 4/4 1504tk 🔶 3/4 1306tk
anthropic/claude-3-5-sonnet-20241022 ✅ 54/56 378tk ✅ 54/56 371tk ✅ 54/56 416tk ✅ 55/56 869tk ✅ 55/56 754tk ❓ N/A ❓ N/A
anthropic/claude-3-5-sonnet-20241022@markdown 🔶 83/118 324tk ✅ 112/118 338tk ✅ 116/118 384tk ✅ 116/118 729tk ✅ 116/118 696tk ❓ N/A ❓ N/A
anthropic/claude-3-5-sonnet-20241022@tool 🔶 60/118 257tk 🔶 46/118 331tk 🔶 60/118 386tk 🔶 57/118 567tk 🔶 60/118 557tk ❓ N/A ❓ N/A
anthropic/claude-3-5-sonnet-20241022@xml ❌ 10/106 293tk 🔶 60/106 307tk 🔶 57/106 369tk ❌ 12/106 631tk 🔶 24/106 658tk ❓ N/A ❓ N/A
anthropic/claude-3-haiku-20240307 ✅ 34/34 388tk ✅ 34/34 375tk ✅ 34/34 432tk ❌ 6/34 781tk 🔶 24/34 903tk 🔶 3/4 670tk ✅ 4/4 1535tk
anthropic/claude-opus-4-1-20250805@markdown 🔶 1/2 393tk ❌ 0/1 197tk ❌ 0/1 204tk ❌ 0/1 190tk ❌ 0/1 186tk ❓ N/A ❓ N/A
anthropic/claude-opus-4-1-20250805@tool ❌ 0/1 181tk ❌ 0/1 196tk ❌ 0/1 202tk ❌ 0/1 188tk ❌ 0/1 184tk ❓ N/A ❓ N/A
anthropic/claude-opus-4-1-20250805@xml 🔶 1/2 372tk ❌ 0/1 197tk ❌ 0/1 205tk ❌ 0/1 191tk ❌ 0/1 187tk ❓ N/A ❓ N/A
anthropic/claude-sonnet-4-20250514@markdown ✅ 1/1 439tk ✅ 1/1 480tk ✅ 1/1 532tk ✅ 1/1 1180tk ✅ 1/1 2006tk ❓ N/A ❓ N/A
anthropic/claude-sonnet-4-20250514@tool ✅ 1/1 283tk ✅ 1/1 320tk ✅ 1/1 471tk ✅ 1/1 1133tk ✅ 1/1 1070tk ❓ N/A ❓ N/A
anthropic/claude-sonnet-4-20250514@xml ✅ 1/1 404tk ✅ 1/1 495tk ✅ 1/1 584tk ✅ 1/1 1223tk ❌ 0/1 1422tk ❓ N/A ❓ N/A
deepseek/deepseek-chat@markdown ✅ 1/1 429tk ❓ N/A ❓ N/A ❓ N/A ❓ N/A ❓ N/A ❓ N/A
deepseek/deepseek-reasoner@markdown ✅ 1/1 742tk ❓ N/A ❓ N/A ❓ N/A ❓ N/A ❓ N/A ❓ N/A
deepseek/deepseek-reasoner@xml ✅ 1/1 680tk ❓ N/A ❓ N/A ❓ N/A ❓ N/A ❓ N/A ❓ N/A
gemini/gemini-1.5-flash-latest ❌ 0/42 28tk ❌ 0/42 28tk ❌ 0/42 28tk ❌ 0/42 28tk ❌ 0/42 28tk ❓ N/A ❓ N/A
gemini/gemini-1.5-flash-latest@markdown ❌ 0/12 29tk ❌ 0/12 29tk ❌ 0/12 29tk ❌ 0/12 29tk ❌ 0/12 29tk ❓ N/A ❓ N/A
gemini/gemini-1.5-flash-latest@tool ❌ 0/12 29tk ❌ 0/12 29tk ❌ 0/12 29tk ❌ 0/12 29tk ❌ 0/12 29tk ❓ N/A ❓ N/A
gemini/gemini-2.5-flash@markdown ✅ 1/1 419tk ✅ 1/1 450tk ✅ 1/1 516tk ✅ 1/1 735tk ❓ N/A ❓ N/A ❓ N/A
gemini/gemini-2.5-flash@xml ✅ 1/1 432tk ✅ 1/1 468tk ✅ 1/1 534tk ✅ 1/1 836tk ❓ N/A ❓ N/A ❓ N/A
groq/moonshotai/kimi-k2-instruct@markdown ✅ 1/1 407tk ✅ 1/1 497tk ✅ 1/1 575tk ✅ 1/1 1079tk ❓ N/A ❓ N/A ❓ N/A
groq/moonshotai/kimi-k2-instruct@xml ✅ 1/1 424tk ✅ 1/1 498tk ✅ 1/1 587tk ✅ 1/1 1150tk ❓ N/A ❓ N/A ❓ N/A
groq/qwen/qwen3-32b@markdown ✅ 1/1 709tk ✅ 1/1 711tk ❌ 0/1 0tk ❌ 0/1 0tk ✅ 1/1 2609tk ❓ N/A ❓ N/A
groq/qwen/qwen3-32b@tool ❌ 0/1 172tk ❌ 0/1 187tk ❌ 0/1 194tk ❌ 0/1 179tk ❌ 0/1 175tk ❓ N/A ❓ N/A
groq/qwen/qwen3-32b@xml ✅ 1/1 1918tk ✅ 1/1 725tk ✅ 1/1 1115tk ❌ 0/1 5171tk ❌ 0/1 4367tk ❓ N/A ❓ N/A
openai/gpt-4-turbo ✅ 3/3 255tk ✅ 3/3 312tk ✅ 3/3 376tk ✅ 3/3 527tk ✅ 4/4 590tk ✅ 4/4 784tk ✅ 6/7 819tk
openai/gpt-4o ✅ 90/91 325tk ✅ 90/91 313tk 🔶 65/91 337tk 🔶 35/91 441tk 🔶 67/91 626tk ✅ 5/5 663tk ✅ 5/5 1253tk
openai/gpt-4o-mini 🔶 60/92 269tk ✅ 91/92 321tk ✅ 91/92 377tk 🔶 62/92 510tk ✅ 80/92 759tk ✅ 6/6 813tk ✅ 6/6 951tk
openai/gpt-4o-mini@markdown ❌ 1/12 217tk ✅ 12/12 302tk ✅ 12/12 361tk 🔶 3/12 440tk ✅ 12/12 823tk ❓ N/A ❓ N/A
openai/gpt-4o-mini@tool ✅ 109/118 279tk ✅ 110/118 301tk ✅ 118/118 391tk ✅ 100/118 573tk ✅ 118/118 665tk ❓ N/A ❓ N/A
openai/gpt-4o@markdown 🔶 69/118 280tk 🔶 72/118 341tk ✅ 113/118 367tk ❌ 12/118 497tk ❌ 21/118 441tk ❓ N/A ❓ N/A
openai/gpt-4o@tool 🔶 43/118 278tk ✅ 96/118 396tk 🔶 93/118 399tk 🔶 69/118 687tk ✅ 118/118 787tk ❓ N/A ❓ N/A
openai/gpt-4o@xml ❌ 8/106 226tk ❌ 16/106 257tk ❌ 4/106 260tk ❌ 1/106 421tk ❌ 2/106 293tk ❓ N/A ❓ N/A
openai/gpt-5-mini@markdown ✅ 1/1 379tk ✅ 1/1 358tk ✅ 1/1 884tk ❌ 0/1 0tk ❌ 0/1 715tk ❓ N/A ❓ N/A
openai/gpt-5-mini@tool ✅ 1/1 286tk ✅ 1/1 294tk ✅ 1/1 1150tk ❌ 0/1 0tk ❌ 0/1 0tk ❓ N/A ❓ N/A
openai/gpt-5-mini@xml ❌ 0/1 331tk ✅ 1/1 386tk ✅ 1/1 922tk ✅ 1/1 1369tk ✅ 1/1 1614tk ❓ N/A ❓ N/A
openai/gpt-5@markdown ✅ 1/1 228tk ✅ 1/1 273tk ✅ 1/1 367tk ✅ 1/1 695tk ❌ 0/1 0tk ❓ N/A ❓ N/A
openai/gpt-5@tool ✅ 1/1 306tk ✅ 1/1 299tk ✅ 1/1 348tk ❌ 0/1 0tk ❌ 0/1 0tk ❓ N/A ❓ N/A
openai/gpt-5@xml ✅ 1/1 237tk ❌ 0/1 218tk ❌ 0/1 217tk ❌ 0/1 678tk ❌ 0/1 255tk ❓ N/A ❓ N/A
openai/o1-mini ✅ 3/3 354tk ✅ 3/3 431tk ✅ 3/3 460tk 🔶 2/3 567tk 🔶 3/5 2222tk 🔶 1/5 1412tk 🔶 2/3 813tk
openai/o1-preview ✅ 2/2 308tk ✅ 2/2 570tk 🔶 1/2 549tk ✅ 2/2 490tk ✅ 3/3 823tk ✅ 1/1 656tk ✅ 1/1 1998tk
google/gemini-flash-1.5 ✅ 2/2 225tk ✅ 2/2 401tk ✅ 2/2 430tk ❌ 0/2 296tk ✅ 1/1 686tk ❌ 0/1 661tk ✅ 1/1 1014tk
google/gemini-pro-1.5 ✅ 1/1 341tk ✅ 1/1 419tk ✅ 1/1 456tk ✅ 1/1 676tk 🔶 2/3 431tk 🔶 1/2 1016tk ✅ 2/2 1308tk
google/gemma-2-27b-it ✅ 1/1 288tk ✅ 1/1 384tk ✅ 1/1 446tk ✅ 1/1 714tk ✅ 1/1 570tk ❌ 0/1 535tk ❌ 0/1 235tk
google/gemma-2-9b-it ❌ 0/2 186tk ✅ 2/2 370tk ✅ 2/2 368tk ❌ 0/2 545tk ✅ 1/1 492tk ❌ 0/1 1730tk ❌ 0/1 352tk
meta-llama/llama-3.1-405b-instruct 🔶 69/91 244tk ✅ 74/91 417tk ✅ 73/91 364tk 🔶 71/91 440tk 🔶 60/91 488tk 🔶 2/5 255tk ❌ 0/5 85tk
meta-llama/llama-3.1-405b-instruct@markdown ✅ 11/12 318tk ✅ 10/12 269tk ✅ 10/12 366tk 🔶 8/12 516tk ✅ 11/12 600tk ❓ N/A ❓ N/A
meta-llama/llama-3.1-405b-instruct@tool ❌ 0/12 184tk ❌ 0/12 198tk ❌ 0/12 214tk ❌ 0/12 190tk ❌ 0/12 189tk ❓ N/A ❓ N/A
meta-llama/llama-3.1-70b-instruct ✅ 5/6 367tk ✅ 5/6 424tk ✅ 6/6 452tk 🔶 2/6 546tk ✅ 5/6 813tk 🔶 3/4 682tk 🔶 2/3 1461tk
meta-llama/llama-3.1-70b-instruct@xml 🔶 61/106 252tk ❌ 8/106 325tk ❌ 19/106 396tk ❌ 7/106 308tk 🔶 40/106 411tk ❓ N/A ❓ N/A
meta-llama/llama-3.1-8b-instruct ✅ 1/1 277tk ✅ 1/1 441tk ❌ 0/1 400tk ❌ 0/1 5095tk ✅ 1/1 2266tk ❓ N/A ❓ N/A
meta-llama/llama-3.2-11b-vision-instruct ✅ 2/2 352tk ✅ 2/2 493tk ❌ 0/2 479tk ✅ 2/2 2643tk ❓ N/A ❓ N/A ❓ N/A
meta-llama/llama-3.2-90b-vision-instruct 🔶 2/4 237tk 🔶 2/4 288tk 🔶 3/4 336tk 🔶 1/4 233tk ❓ N/A ❓ N/A ❓ N/A
mistralai/magistral-medium-2506@markdown ✅ 1/1 531tk ✅ 1/1 569tk ✅ 1/1 666tk ✅ 1/1 1106tk ❓ N/A ❓ N/A ❓ N/A
mistralai/magistral-medium-2506@tool ❌ 0/1 465tk ❌ 0/1 604tk ❌ 0/1 0tk ❌ 0/1 0tk ❓ N/A ❓ N/A ❓ N/A
mistralai/magistral-medium-2506@xml ✅ 1/1 516tk ✅ 1/1 568tk ❌ 0/1 552tk ✅ 1/1 1075tk ❓ N/A ❓ N/A ❓ N/A
moonshotai/kimi-k2-0905@markdown ✅ 2/2 464tk ✅ 2/2 590tk ✅ 2/2 613tk 🔶 1/2 650tk ❓ N/A ❓ N/A ❓ N/A
moonshotai/kimi-k2-0905@tool ❌ 0/1 397tk ❌ 0/1 483tk ❌ 0/1 592tk ✅ 1/1 990tk ❓ N/A ❓ N/A ❓ N/A
moonshotai/kimi-k2-0905@xml ✅ 1/1 441tk ✅ 1/1 563tk ✅ 1/1 848tk ❌ 0/1 598tk ❓ N/A ❓ N/A ❓ N/A
nousresearch/hermes-2-pro-llama-3-8b ✅ 1/1 341tk ❌ 0/1 4274tk ❌ 0/1 3760tk ❌ 0/1 659tk ❓ N/A ❓ N/A ❓ N/A
nousresearch/hermes-3-llama-3.1-405b ✅ 2/2 317tk ✅ 2/2 420tk ✅ 2/2 325tk ✅ 2/2 410tk ✅ 1/1 821tk ✅ 1/1 758tk ✅ 1/1 1039tk
nousresearch/hermes-3-llama-3.1-70b ❌ 0/2 173tk ❌ 0/2 187tk ❌ 0/2 202tk ❌ 0/2 177tk ❓ N/A ❓ N/A ❓ N/A
nousresearch/hermes-4-70b@markdown ❌ 0/1 439tk ✅ 1/1 612tk ✅ 1/1 1235tk ❌ 0/1 0tk ❓ N/A ❓ N/A ❓ N/A
nousresearch/hermes-4-70b@tool ✅ 1/1 476tk ✅ 1/1 536tk ❌ 0/1 1310tk ❌ 0/1 0tk ❓ N/A ❓ N/A ❓ N/A
nousresearch/hermes-4-70b@xml ✅ 1/1 466tk ✅ 1/1 516tk ✅ 1/1 631tk ❌ 0/1 1255tk ❓ N/A ❓ N/A ❓ N/A
qwen/qwen3-max@markdown ✅ 1/1 422tk ✅ 1/1 492tk 🔶 1/2 615tk ✅ 1/1 949tk ❓ N/A ❓ N/A ❓ N/A
qwen/qwen3-max@tool ✅ 1/1 436tk ✅ 1/1 519tk ✅ 1/1 546tk ✅ 1/1 663tk ❓ N/A ❓ N/A ❓ N/A
qwen/qwen3-max@xml ✅ 1/1 437tk ✅ 1/1 530tk ✅ 1/1 580tk ✅ 1/1 901tk ❓ N/A ❓ N/A ❓ N/A
x-ai/grok-4-fast:free@markdown ✅ 1/1 561tk 🔶 1/2 760tk ✅ 1/1 1326tk ✅ 1/1 1016tk ❓ N/A ❓ N/A ❓ N/A
x-ai/grok-code-fast-1@markdown ✅ 1/1 661tk ❌ 0/1 1385tk ✅ 1/1 829tk ✅ 1/1 955tk ❓ N/A ❓ N/A ❓ N/A
x-ai/grok-code-fast-1@tool ✅ 1/1 663tk ❌ 0/1 2590tk ❌ 0/1 1415tk ✅ 1/1 1807tk ❓ N/A ❓ N/A ❓ N/A
x-ai/grok-code-fast-1@xml ❌ 0/1 485tk ❌ 0/1 1652tk ✅ 1/1 759tk ✅ 1/1 1112tk ❓ N/A ❓ N/A ❓ N/A
We are working on making the evals more robust, informative, and challenging.
Other evals#
We have considered running gptme on other evals such as SWE-Bench, but have not finished it (see PR #142).
If you are interested in running gptme on other evals, drop a comment in the issues!