Compare · ModelsLive · 2 picked · head to head
GPT-5 Nano vs gpt-oss-20b
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
GPT-5 Nano wins on 8/12 benchmarks
GPT-5 Nano wins 8 of 12 shared benchmarks. Leads in knowledge · language · reasoning.
Category leads
knowledge·GPT-5 Nanogeneral·gpt-oss-20blanguage·GPT-5 Nanomath·gpt-oss-20breasoning·GPT-5 Nanocoding·GPT-5 Nano
Hype vs Reality
Attention vs performance
GPT-5 Nano
#193 by perf·#4 by attention
gpt-oss-20b
#97 by perf·no signal
Best value
gpt-oss-20b
5.6x better value than GPT-5 Nano
GPT-5 Nano
174.2 pts/$
$0.23/M
gpt-oss-20b
983.3 pts/$
$0.05/M
Vendor risk
Who is behind the model
OpenAI
$840.0B·Tier 1
OpenAI
$840.0B·Tier 1
Head to head
12 benchmarks · 2 models
GPT-5 Nanogpt-oss-20b
Chess Puzzles
GPT-5 Nano leads by +23.2
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
GPT-5 Nano
23.2
gpt-oss-20b
0.0
Dtbench
gpt-oss-20b leads by +8.9
GPT-5 Nano
37.8
gpt-oss-20b
46.7
GPQA diamond
GPT-5 Nano leads by +11.5
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
GPT-5 Nano
59.3
gpt-oss-20b
47.7
HELM · GPQA
GPT-5 Nano leads by +8.5
GPT-5 Nano
67.9
gpt-oss-20b
59.4
HELM · IFEval
GPT-5 Nano leads by +20.0
GPT-5 Nano
93.2
gpt-oss-20b
73.2
HELM · MMLU-Pro
GPT-5 Nano leads by +3.8
GPT-5 Nano
77.8
gpt-oss-20b
74.0
HELM · Omni-MATH
gpt-oss-20b leads by +1.8
GPT-5 Nano
54.7
gpt-oss-20b
56.5
HELM · WildBench
GPT-5 Nano leads by +6.9
GPT-5 Nano
80.6
gpt-oss-20b
73.7
Lmca
gpt-oss-20b leads by +7.7
GPT-5 Nano
9.3
gpt-oss-20b
17.1
OTIS Mock AIME 2024-2025
GPT-5 Nano leads by +15.9
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
GPT-5 Nano
81.1
gpt-oss-20b
65.2
Terminal Bench
GPT-5 Nano leads by +18.4
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 Nano
21.8
gpt-oss-20b
3.4
WeirdML
gpt-oss-20b leads by +2.9
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
GPT-5 Nano
38.1
gpt-oss-20b
40.9
Full benchmark table
| Benchmark | GPT-5 Nano | gpt-oss-20b |
|---|---|---|
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 23.2 | 0.0 |
Dtbench | 37.8 | 46.7 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 59.3 | 47.7 |
HELM · GPQA | 67.9 | 59.4 |
HELM · IFEval | 93.2 | 73.2 |
HELM · MMLU-Pro | 77.8 | 74.0 |
HELM · Omni-MATH | 54.7 | 56.5 |
HELM · WildBench | 80.6 | 73.7 |
Lmca | 9.3 | 17.1 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 81.1 | 65.2 |
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. | 21.8 | 3.4 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 38.1 | 40.9 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.05 | $0.40 | 400K tokens (~200 books) | $1.38 | |
| $0.02 | $0.09 | 131K tokens (~66 books) | $0.36 |
People also compared