Request a pilot
‹ PublicAI Index

The LLM benchmark aggregator.

GPT-3.5 Turbo

OpenAI

Strongest in Safe-prompt compliance (#60 of 82), weakest in Fairness (#300 of 300). Above par in 1 of 11 scopes. Among the models it meets almost everywhere, it finishes behind GPT-5.1 and GPT-5 and ahead of Mistral Instruct v0.3 (7B) and Zephyr 7B Beta.

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Reasoning38.7#158/178
Reasoning31.8#140/140
Human preference38.1#274/342
Human preference38.1#274/342
Safety44.6#292/337
Safe-prompt compliance49.3#60/82
Jailbreak resistance52.5#151/272
Secure code47.3#184/274
Harm refusal46#226/300
Toxicity avoidance43.6#235/272
Fairness33.3#300/300

§ 2 · Head to head

What it beats, and what beats it

The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for GPT-3.5 Turbo, left for the other.

Smollm2 1.7B Instructwon 7 · lost 1
Gemma 7B Itwon 7 · lost 1
Nous Hermes 2 Mixtral 8x7b Dpowon 7 · lost 1
OLMoE 1B-7B Instructwon 6 · lost 1
Mistral Instruct v0.1 (7B)won 6 · lost 1
DBRX Instructwon 7 · lost 2
Gemma 2B Itwon 6 · lost 2
Magistral Smallwon 5 · lost 3
Zephyr 7B Betawon 5 · lost 3
Mistral Instruct v0.3 (7B)won 4 · lost 3
Grok Build 0.1won 4 · lost 4
Mistral 7B Instruct V0.2won 4 · lost 4
Qwen1.5 14B Chatwon 4 · lost 4
Gemma 2 2B Itwon 4 · lost 4
Grok 3won 5 · lost 5
DeepSeek LLM 67B Chatwon 4 · lost 5
OLMo 2 7B Instructwon 3 · lost 4
Jamba 1.6 Largewon 3 · lost 4
Grok 4.1 Fastwon 4 · lost 6
Llama 3.2 1B Instructwon 3 · lost 5
Llama 3.2 3B Instructwon 3 · lost 5
Starling Lm 7B Betawon 3 · lost 5
Granite 3.1 2B Instructwon 3 · lost 5
Llama 3.1 8B Instructwon 3 · lost 5
Jamba 1.5 Miniwon 3 · lost 5
Qwen3 30B A3B Instructwon 3 · lost 5
Mistral Largewon 3 · lost 6
Command Awon 3 · lost 6
Mistral Smallwon 3 · lost 6
GPT OSS 20Bwon 3 · lost 7
Llama 3.1 Instruct Turbo (70B)won 2 · lost 5
IBM Granite 4.0 Microwon 2 · lost 5
Qwen 2.5 Instruct Turbo (7B)won 2 · lost 5
Jamba 1.6 Miniwon 2 · lost 5
Llama 3.3 Instruct Turbo (70B)won 2 · lost 5
Ling 3.0 Flashwon 2 · lost 5
Gemini 1.5 Prowon 3 · lost 8
GPT OSS 120Bwon 3 · lost 8
Grok 4won 3 · lost 8
DeepSeek R1won 3 · lost 8
Nemotron 3 Ultra 550B A55Bwon 2 · lost 6
Phi 3 Mini 128k Instructwon 2 · lost 6
Phi 3 Mini 4k Instructwon 2 · lost 6
Granite 3.0 2B Instructwon 2 · lost 6
Phi 3 Small 8k Instructwon 2 · lost 6
Granite 3.0 8B Instructwon 2 · lost 6
Phi 3 Medium 4k Instructwon 2 · lost 6
Gemma 2 9B Itwon 2 · lost 6
Gemma 3 4B Itwon 2 · lost 6
NVIDIA Nemotron 3 Super 120B A12Bwon 2 · lost 6
Claude 3.5 Haikuwon 2 · lost 7
Gemini 2.0 Flash Litewon 2 · lost 7
GLM-4.5won 2 · lost 7
GLM-4.6won 2 · lost 7
Magistral Mediumwon 2 · lost 8
Gemini 1.5 Flashwon 2 · lost 8
DeepSeek V3.2won 2 · lost 8
MiniMax M3won 2 · lost 8
Kimi K2.6won 2 · lost 8
Qwen3.7 Maxwon 2 · lost 8
GLM-5.2won 2 · lost 8
Gemini 3.5 Flashwon 2 · lost 8
Gemini 3.6 Flashwon 2 · lost 8
Command R Pluswon 2 · lost 9
Llama 4 Scout Instructwon 2 · lost 9
DeepSeek V3won 2 · lost 9
Claude Haiku 4.5won 2 · lost 9
Kimi K2won 2 · lost 9
Llama 3.1 Instruct Turbo (8B)won 1 · lost 6
Llama 3.1 Instruct Turbo (405B)won 1 · lost 6
Gemma 4 E4Bwon 1 · lost 6
Gemma 4 12Bwon 1 · lost 6
Kimi K2.7 Codewon 1 · lost 7
Kimi K2 Thinkingwon 1 · lost 7
Granite 3.1 8B Instructwon 1 · lost 7
Mistral Small 24B Instructwon 1 · lost 7
Jamba 1.5 Largewon 1 · lost 7
Nova Prowon 1 · lost 7
QwQ 32Bwon 1 · lost 7
Qwen3.5 35B A3Bwon 1 · lost 7
Qwen3.5 27Bwon 1 · lost 7
Qwen3 Vl 235B A22B Instructwon 1 · lost 7
Qwen3.5 122B A10Bwon 1 · lost 7
Gemma 4 26B A4Bwon 1 · lost 7
Qwen3.5 397B A17Bwon 1 · lost 7
Claude Opus 4.6won 1 · lost 7
Llama 3 8B Instructwon 1 · lost 8
Claude 3 Haikuwon 1 · lost 8
Llama 3 70B Instructwon 1 · lost 8
Claude 3 Sonnetwon 1 · lost 8
Gemma 4 31Bwon 1 · lost 8
Mimo V2.6 Flashwon 1 · lost 8
Mimo V2.5 Prowon 1 · lost 8
DeepSeek V4 Flashwon 1 · lost 9
GPT-6 Lunawon 1 · lost 9
Grok 4.3won 1 · lost 9
Kimi K2.5won 1 · lost 9
GPT-6 Solwon 1 · lost 9
GLM-5.1won 1 · lost 9
GLM-5.3 Flashwon 1 · lost 9
Claude Opus 5won 1 · lost 9
Gemini 3.8 Flashwon 1 · lost 9
GPT-4o Miniwon 1 · lost 10
Claude 3 Opuswon 1 · lost 10
GPT-4.1 Nanowon 1 · lost 10
Llama 4 Maverick Instructwon 1 · lost 10
GPT-4owon 1 · lost 10
Gemini 2.0 Flashwon 1 · lost 10
GPT-4.1 Miniwon 1 · lost 10
Claude 3.7 Sonnetwon 1 · lost 10
Gemini 2.5 Flashwon 1 · lost 10
Qwen3 235B A22B Instructwon 1 · lost 10
Gemini 2.5 Prowon 1 · lost 10
Claude Sonnet 4.5won 1 · lost 10
Qwen 2.5 Instruct Turbo (72B)won 0 · lost 7
Gemini 2.5 Flash Litewon 0 · lost 7
Llama 3.1 Tulu 3 8Bwon 0 · lost 8
C4ai Aya Expanse 8Bwon 0 · lost 8
Nova Microwon 0 · lost 8
Phi 4won 0 · lost 8
Nova Litewon 0 · lost 8
C4ai Aya Expanse 32Bwon 0 · lost 8
GPT-4won 0 · lost 8
Gemma 2 27B Itwon 0 · lost 8
Qwen2.5 72B Instructwon 0 · lost 8
Qwen Maxwon 0 · lost 8
Gemma 3 12B Itwon 0 · lost 8
Gemma 3 27B Itwon 0 · lost 8
GPT-4.5won 0 · lost 8
Hy3won 0 · lost 8
Claude Sonnet 4.6won 0 · lost 8
Gemini 3 Prowon 0 · lost 8
Qwen2 72B Instructwon 0 · lost 9
MiniMax M2.7won 0 · lost 9
Mimo V2.6 Prowon 0 · lost 9
MiniMax M2won 0 · lost 10
GPT-5.2won 0 · lost 10
Grok 4.7won 0 · lost 10
Inklingwon 0 · lost 10
Grok 4.6won 0 · lost 10
GPT-5.6 Lunawon 0 · lost 10
Claude Sonnet 5won 0 · lost 10
DeepSeek V4 Prowon 0 · lost 10
Grok 4.5won 0 · lost 10
GPT-5.6 Terrawon 0 · lost 10
GPT-5.4won 0 · lost 10
Qwen3.8 Maxwon 0 · lost 10
Claude Opus 4.8won 0 · lost 10
GPT-5.5won 0 · lost 10
GPT-5.6 Solwon 0 · lost 10
Muse Spark 1.1won 0 · lost 10
Claude Opus 4.7won 0 · lost 10
Claude Opus 5.5won 0 · lost 10
GPT-4 Turbowon 0 · lost 11
O1 Miniwon 0 · lost 11
GPT-5 Nanowon 0 · lost 11
O3 Miniwon 0 · lost 11
Claude 3.5 Sonnetwon 0 · lost 11
GPT-5 Miniwon 0 · lost 11
O4 Miniwon 0 · lost 11
Claude Sonnet 4won 0 · lost 11
O1won 0 · lost 11
GPT-4.1won 0 · lost 11
Claude Opus 4won 0 · lost 11
O3won 0 · lost 11
GPT-5won 0 · lost 11
GPT-5.1won 0 · lost 11

§ 3 · Sources

Where the numbers come from

4 publications, 13 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-09-26
LMArena Text 1204
SimpleBench ↗1 measureread 2026-09-26
SimpleBench 8%
HELM Safety ↗5 measuresread 2026-09-26
HELM Safety · HarmBench 68.6%HELM Safety · SimpleSafetyTests 92.3%HELM Safety · Anthropic Red Team 98%HELM Safety · BBQ 64.9%HELM Safety · XSTest 93.8%
Enkrypt AI Safety Leaderboard ↗6 measuresread 2026-09-26
Enkrypt · Jailbreak risk 10.7%Enkrypt · Harmful content risk 62.8%Enkrypt · CBRN risk 12.8%Enkrypt · Toxicity risk 10.8%Enkrypt · Bias risk 89.9%Enkrypt · Insecure code risk 39.1%

Badge

PublicAI Index badge for GPT-3.5 Turbo[![PublicAI Index](https://publicai.io/model-index/badge?model=gpt-3-5-turbo)](https://publicai.io/model-index/m/gpt-3-5-turbo)