whatmodel.sh / GPT-3.5 Turbo (0301)
GPT-3.5 Turbo (0301)
↗ share
+ contribute
GPT-3.5 Turbo (0301) is a closed-weights language model from OpenAI, released March 2023. Licence: proprietary.
Catalogue identity notes
Comparison-only. Dated API snapshot gpt-3.5-turbo-0301, the first ChatGPT API model. Listed as shut down on OpenAI's deprecations page, so no current API. Release month from the snapshot date in the model name.
$ how it compares
The LLMs ranked either side of GPT-3.5 Turbo (0301), out of 828 with a score, then more from OpenAI.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 500 |  Granite 4.0 Micro 3B | 46±1 | local | ||
| 501 |  Command R (08-2024) | 46±1 | api $0.26/M +local | ||
| 502 | GPT-3.5 Turbo (0301) | 46±1 | comparison only | ||
| 503 |  Nova Micro | 46±2 | api $0.061/M | ||
| 504 |  T5 Base | 46±9 | local | ||
| more from OpenAI | |||||
| 1 | GPT-6 Astra | 87±1 | api $20.00/M | ||
| 3 | GPT-6.1 Sol | 85±1 | api $4.00/M | ||
| 7 | GPT-6 Sol | 83±1 | api $4.00/M | ||
$ what it's good at
coding▁▁▁▃▅▇█▇▅▄▃▁ #344 of 524 · 64% of its score · 4 tests
Beats Qwen3-VL 32B Instruct, Solar Open 100B (102B-A12B), GLM-4.7-Flash here, though they rank higher overall. Trails DeepSeek LLM 67B Chat, WizardLM 70B V1.0, Yi 1.5 9B Chat here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| HumanEval | 73.2% ✎ | ▁·▁▁▃▄▆▅█▃▁▁ beats 65/99 | 30% |
| HumanEval+ | 66.5% ✎ | ▁·▁▁▁▃▄▄▄█▃▁ beats 63/99 | 24% |
| Aider code editing (133 Exercism Python exercises, two attempts) | 57.9% ✎ | ▁·▁▂▄▁▆▄█▆▂▂ beats 27/62 | 7.0% |
| Tulu 2: CodexEval pass@10 | 88.4% ✎ | ▃▃▃·▆▃▃█▃▂▂▅ beats 25/25 | 2.1% |
chat & writing▁▁▁▃▅▄▆█▇▇▇▃ #383 of 463 · 24% of its score · 3 tests
Beats Gemma 4 E4B, OpenThinker3 7B, Gemma 4 E2B here, though they rank higher overall. Trails LFM2 350M, Gemma 3 1B, Flan-T5 XXL here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| AlpacaEval 2 length-controlled | 18.1% ✎ | ▁▂▂▃▆▆█▆▄▄▁▁ beats 48/122 | 14% |
| Arena-Hard (v0.1, GPT-4-1106 judge) | 18.1% ✎ | ▂▁▁▅█▅▄▃▃▅▃▁ beats 12/60 | 9.5% |
| AlpacaEval 1.0 win rate (vs text-davinci-003, GPT-4 judge) | 89.4% ✎ | ▂▂·▆·▇▃▆█▆▄▂ beats 30/45 | 0.6% |
math▁▂▃▅▇▅▅█▄▁▁▁ #252 of 462 * · 1.9% of its score · 1 test
Beats GPT-4o (2024-08-06), Command A (03-2025), GPT-4o (2024-11-20) here, though they rank higher overall. Trails Olmo 3 7B Instruct here, though they rank lower overall.
other tests4 not in a task · 11% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| BIG-Bench Hard | 66.1% ✎ | ▁·····▂█▂▅·▁ beats 22/25 | 6.0% |
| MMLU | 67.9% ✎ | ▁▁▃▃▄█▄▁█··▁ beats 24/25 | 2.2% |
| EQ-Bench v1 (60 questions, paper run) | 47.6% ✎ | ▂··▄·▂▆▄▃█·▂ beats 9/14 | 1.7% |
| Tulu 2: TyDiQA gold passage, 1-shot F1 | 51.9% ✎ | ▂··▁▃▂▂▂█▅▁▂ beats 14/25 | 0.7% |
$ where to run it
No verified API or download recorded yet · add one
$ how the score was worked out · EvalPlus / EvalPerf published… 55%, AlpacaEval 1.0 and 2.0 leader… 14%, Tulu 2 — base and instruction… 13% and 3 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 EvalPlus / EvalPerf published evaluation runs | 2 | 55% |
 AlpacaEval 1.0 and 2.0 leaderboards (tatsu-lab/alpaca_eval repository) | 2 | 14% |
 Tulu 2 — base and instruction-tuned evaluations | 5 | 13% |
| Arena-Hard-Auto v0.1 leaderboard (repository README, pre-v2) | 1 | 9% |
| Aider code editing benchmark runs (old edit leaderboard) | 1 | 7% |
| EQ-Bench: An Emotional Intelligence Benchmark for Large Language Models — authors' own runs | 1 | 2% |
| prior · pull toward the middle | 0% | |
| weighted average | = 46.0 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json