OpenAIproprietary
Released: 2026-06-01

OpenAI: GPT-3.5 Turbo Instruct

This model is a variant of GPT-3.5 Turbo tuned for instructional prompts and omitting chat-related optimizations. Training data: up to Sep 2021.

Intelligence Index
85.0/ 100
Calibrated multi-domain composite
Throughput Speed
85tok/sec
Output streaming throughput
Time To First Token
280ms
Initial chunk server latency
LiveBench Contam-Free
74.8%
September 2026 suite score

LiveBench Multi-Domain Evaluation Breakdown

Monthly refreshed contamination-free test sets
Logical Reasoning78.2%
Coding & Repo Repair76.5%
Mathematics (AIME)75.7%
Data Analysis & Tables72.3%
Language & Comprehension74.0%
Instruction Following74.8%

Sourced Benchmark Evaluations (51)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-3Active53.6%
93
2026-07-06independentSource ↗
ARC-AGI-2Nearing Saturation64.4%
75
2026-06-29independentSource ↗
FrontierMathActive22.1%
98
2026-06-26independentSource ↗
PutnamBenchActive41.1%
90
2026-06-25independentSource ↗
Omni-MATHDeprecated58.5%
10
2026-06-23independentSource ↗
SWE-LancerDeprecated54.9%
20
2026-06-23independentSource ↗
SoohakActive60.1%
98
2026-06-21independentSource ↗
FrontierCodeActive46.9%
98
2026-06-21independentSource ↗
AIMENearing Saturation75.1%
75
2026-06-19independentSource ↗
Humanity's Last ExamActive36.3%
93
2026-06-19independentSource ↗
StructFlowBenchNearing Saturation63.6%
75
2026-06-19independentSource ↗
OlympiadBenchActive64.9%
80
2026-06-18independentSource ↗
MathArenaActive63.3%
99
2026-06-17independentSource ↗
FireBenchActive62.8%
95
2026-06-17independentSource ↗
Terminal-BenchActive56.6%
80
2026-06-17independentSource ↗
CodeContestsDeprecated52.5%
10
2026-06-16independentSource ↗
SimpleQAActive44.2%
90
2026-06-16independentSource ↗
SWE-bench ProActive50.1%
90
2026-06-15independentSource ↗
ComplexBenchNearing Saturation61.3%
80
2026-06-15independentSource ↗
MultiPL-EDeprecated74.3%
10
2026-06-14independentSource ↗
AgentIFNearing Saturation50.9%
80
2026-06-13independentSource ↗
MATHSaturated77.5%
25
2026-06-13independentSource ↗
LiveCodeBenchSaturated66.3%
39
2026-06-13independentSource ↗
MercuryDeprecated71.1%
10
2026-06-12independentSource ↗
MMLU-ProNearing Saturation78.2%
75
2026-06-12independentSource ↗
CyberSecEvalDeprecated72.7%
10
2026-06-11independentSource ↗
TabMWPSaturated72.8%
25
2026-06-11independentSource ↗
SWE-bench VerifiedSaturated63%
25
2026-06-10independentSource ↗
MMLU-ReduxSaturated83.3%
25
2026-06-10independentSource ↗
SWE-benchDeprecated54.9%
10
2026-06-09independentSource ↗
MGSMNearing Saturation75.1%
65
2026-06-09independentSource ↗
TruthfulQASaturated74.8%
25
2026-06-08independentSource ↗
AGIEvalSaturated71%
25
2026-06-08independentSource ↗
SuperGLUESaturated84.2%
25
2026-06-07independentSource ↗
GSM-HardDeprecated75.9%
10
2026-06-07independentSource ↗
MT-BenchNearing Saturation7.7%
75
2026-06-06independentSource ↗
CodeXGLUEDeprecated76%
23
2026-06-06independentSource ↗
BIG-Bench HardSaturated80.9%
25
2026-06-06independentSource ↗
MBPP+Nearing Saturation82.4%
65
2026-06-05independentSource ↗
IFEvalDeprecated73.3%
25
2026-06-05independentSource ↗
WinoGrandeSaturated85.8%
28
2026-06-05independentSource ↗
GPQANearing Saturation74.8%
78
2026-06-05independentSource ↗
HumanEval+Saturated87.3%
25
2026-06-04independentSource ↗
CommonsenseQASaturated86.6%
28
2026-06-04independentSource ↗
MBPPSaturated85.6%
25
2026-06-03independentSource ↗
GSM8KSaturated85.4%
25
2026-06-03independentSource ↗
GLUESaturated89.1%
28
2026-06-03independentSource ↗
MMLUSaturated86.7%
25
2026-06-02independentSource ↗
ARC-AGISaturated77.6%
25
2026-06-02independentSource ↗
HumanEvalSaturated92.1%
25
2026-06-02independentSource ↗
LiveBench (Continuous)Active74.8%
94
2025-02-01independentSource ↗
OpenAI Compatible API ExecutionAPI Ready
from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="openai-gpt-3-5-turbo-instruct",
    messages=[{"role": "user", "content": "Evaluate multi-step logic problem"}],
    temperature=0.2
)
print(response.choices[0].message.content)

Evaluation & Sourcing Notes

Scores listed for OpenAI: GPT-3.5 Turbo Instruct represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.