Compare · ModelsLive · 2 picked · head to head
GPT-5 Mini vs gpt-oss-120b
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-5 Mini wins on 23/25 benchmarks
GPT-5 Mini wins 23 of 25 shared benchmarks. Leads in knowledge · general · language.
Category leads
knowledge·GPT-5 Minigeneral·GPT-5 Minilanguage·GPT-5 Minimath·GPT-5 Minireasoning·GPT-5 Minicoding·GPT-5 Mini
Hype vs Reality
Attention vs performance
GPT-5 Mini
#108 by perf·#4 by attention
gpt-oss-120b
#127 by perf·no signal
Best value
gpt-oss-120b
10.4x better value than GPT-5 Mini
GPT-5 Mini
45.8 pts/$
$1.13/M
gpt-oss-120b
475.4 pts/$
$0.10/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
25 benchmarks · 2 models
GPT-5 Minigpt-oss-120b
Chess Puzzles
GPT-5 Mini leads by +10.5
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
GPT-5 Mini
26.4
gpt-oss-120b
15.8
Dtbench
GPT-5 Mini leads by +7.1
GPT-5 Mini
67.5
gpt-oss-120b
60.5
Fiction.LiveBench
GPT-5 Mini leads by +25.0
Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination.
GPT-5 Mini
69.4
gpt-oss-120b
44.4
GPQA diamond
gpt-oss-120b leads by +1.0
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-5 Mini
66.7
gpt-oss-120b
67.7
HELM · GPQA
GPT-5 Mini leads by +7.2
GPT-5 Mini
75.6
gpt-oss-120b
68.4
HELM · IFEval
GPT-5 Mini leads by +9.1
GPT-5 Mini
92.7
gpt-oss-120b
83.6
HELM · MMLU-Pro
GPT-5 Mini leads by +4.0
GPT-5 Mini
83.5
gpt-oss-120b
79.5
HELM · Omni-MATH
GPT-5 Mini leads by +3.4
GPT-5 Mini
72.2
gpt-oss-120b
68.8
HELM · WildBench
GPT-5 Mini leads by +1.0
GPT-5 Mini
85.5
gpt-oss-120b
84.5
Lech Mazur Writing
GPT-5 Mini leads by +6.0
Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication.
GPT-5 Mini
83.1
gpt-oss-120b
77.1
LiveBench · Agentic Coding
GPT-5 Mini leads by +18.3
GPT-5 Mini
35.0
gpt-oss-120b
16.7
LiveBench · Coding
GPT-5 Mini leads by +15.9
GPT-5 Mini
76.1
gpt-oss-120b
60.2
LiveBench · Data Analysis
GPT-5 Mini leads by +10.8
GPT-5 Mini
49.6
gpt-oss-120b
38.8
LiveBench · If
GPT-5 Mini leads by +13.9
GPT-5 Mini
64.2
gpt-oss-120b
50.3
LiveBench · Language
GPT-5 Mini leads by +20.6
GPT-5 Mini
69.2
gpt-oss-120b
48.6
LiveBench · Mathematics
GPT-5 Mini leads by +5.5
GPT-5 Mini
74.4
gpt-oss-120b
68.9
LiveBench · Overall
GPT-5 Mini leads by +14.9
GPT-5 Mini
61.0
gpt-oss-120b
46.1
LiveBench · Reasoning
GPT-5 Mini leads by +19.4
GPT-5 Mini
58.6
gpt-oss-120b
39.2
Lmca
GPT-5 Mini leads by +14.2
GPT-5 Mini
40.2
gpt-oss-120b
26.1
Mystery Game Puzzles
GPT-5 Mini leads by +0.9
GPT-5 Mini
0.9
gpt-oss-120b
0.0
OTIS Mock AIME 2024-2025
gpt-oss-120b leads by +2.2
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-5 Mini
86.7
gpt-oss-120b
88.9
SimpleQA Verified
GPT-5 Mini leads by +7.7
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
GPT-5 Mini
21.6
gpt-oss-120b
13.9
SWE-Bench Verified (Bash Only)
GPT-5 Mini leads by +33.8
SWE-Bench Verified (Bash Only) · a curated subset of SWE-bench where models fix real Python repository bugs using only bash commands, no agent frameworks.
GPT-5 Mini
59.8
gpt-oss-120b
26.0
Terminal Bench
GPT-5 Mini leads by +16.1
Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency. Claude Opus 4.7 scores 69.4%, demonstrating significant agentic terminal competence.
GPT-5 Mini
34.8
gpt-oss-120b
18.7
WeirdML
GPT-5 Mini leads by +4.5
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-5 Mini
52.7
gpt-oss-120b
48.2
Full benchmark table
| Benchmark | GPT-5 Mini | gpt-oss-120b |
|---|---|---|
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 26.4 | 15.8 |
Dtbench | 67.5 | 60.5 |
Fiction.LiveBench Fiction.LiveBench · a continuously updated benchmark using recently published fiction to test reading comprehension and reasoning, preventing data contamination. | 69.4 | 44.4 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 66.7 | 67.7 |
HELM · GPQA | 75.6 | 68.4 |
HELM · IFEval | 92.7 | 83.6 |
HELM · MMLU-Pro | 83.5 | 79.5 |
HELM · Omni-MATH | 72.2 | 68.8 |
HELM · WildBench | 85.5 | 84.5 |
Lech Mazur Writing Lech Mazur Writing · evaluates creative writing ability, assessing prose quality, narrative coherence, and stylistic sophistication. | 83.1 | 77.1 |
LiveBench · Agentic Coding | 35.0 | 16.7 |
LiveBench · Coding | 76.1 | 60.2 |
LiveBench · Data Analysis | 49.6 | 38.8 |
LiveBench · If | 64.2 | 50.3 |
LiveBench · Language | 69.2 | 48.6 |
LiveBench · Mathematics | 74.4 | 68.9 |
LiveBench · Overall | 61.0 | 46.1 |
LiveBench · Reasoning | 58.6 | 39.2 |
Lmca | 40.2 | 26.1 |
Mystery Game Puzzles | 0.9 | 0.0 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 86.7 | 88.9 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 21.6 | 13.9 |
SWE-Bench Verified (Bash Only) SWE-Bench Verified (Bash Only) · a curated subset of SWE-bench where models fix real Python repository bugs using only bash commands, no agent frameworks. | 59.8 | 26.0 |
Terminal Bench Terminal-Bench 2.0 · evaluates AI agents on real terminal-based coding tasks · writing scripts, debugging, running tests, and managing projects entirely through command-line interaction. Tests both code quality and terminal fluency. Claude Opus 4.7 scores 69.4%, demonstrating significant agentic terminal competence. | 34.8 | 18.7 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 52.7 | 48.2 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.25 | $2.00 | 400K tokens (~200 books) | $6.88 | |
| $0.04 | $0.17 | 131K tokens (~66 books) | $0.70 |